Defining and Measuring Crime

Some Notes that have been hanging around on my hard drive doing nothing useful that I’ve finally got around to posting. There are plenty more where these came from but whether or not I’ll ever get around to digging them out is anyone’s guess.

Defining Crime and Deviance


‘To deviate’ means ‘to stray from the path’ and the path here is behaviour that a society considers normal – something that is always a matter for debate. In our society, for example, it’s normal to maintain a circle of space around our body extending roughly 60 cm, and we feel uncomfortable if people enter this ‘personal space’ uninvited. In other cultures, concepts of personal space are different – in Argentina, for example, personal space can be so small as to be almost non-existent. This example illustrates that:

  • all cultures develop ideas about ‘normal behaviour’
  • the rules – or norms – governing normal behaviour can be different from culture to culture (and frequently within the same culture at different times)

Deviance, therefore, refers to actions that deviate from the norm – the ‘underlying rules of social interaction’ that take two basic forms:

  • Formal norms involve laws and organisational rules that represent official standards of behaviour. They are usually written, formally policed and punishment (or ‘negative sanctions’) for deviance is clearly specified as part of the rule. If you break the law, for example, you risk being arrested and imprisoned. Unlike laws, organisational rules apply to a particular group or organisation, rather than society as a whole. School pupils, for example, may have to wear a uniform, but this rule doesn’t apply to their parents.
  • Informal norms are unwritten, informally policed and carry informal punishments that vary from group to group. This means the same behaviour may be seen differently depending on its context; swearing when with a group of friends may be considered acceptable behaviour, while swearing in front of your parents may be considered quite differently.


Crime, or behaviour that breaks legal rules (formal norms), is a subset of deviance in the sense that while all forms of crime are deviant, not all forms of deviance are criminal; forgetting your best friend’s birthday, for example, may be deviant, but it’s not a criminal offence.

The social construction of crime and deviance

The example of personal space suggests that concepts of deviance vary in both time and space. At different times in a society’s historical development, for example, the same behaviour may be seen as:

  • criminal and deviant
  • not criminal, but deviant
  • neither criminal nor deviant

An example here is homosexuality: criminal and deviant until 1967, for most people in contemporary Britain this behaviour is seen as neither criminal nor necessarily deviant.

In terms of cultural variations, different societies often develop different behavioural rules. Adults drinking alcohol, for example, is neither criminal nor deviant in Western societies (although there are age variations – 18 in Britain, 21 in the USA); the same behaviour is illegal in countries such as Saudi Arabia.


These examples show deviance is both:

  • relative (its meaning changes over time and between societies), and
  • socially constructed – there is nothing inherently deviant (deviant in all societies and at all times) about any form of behaviour. What matters is the meaning of that behaviour. For example, even an act as extreme as killing someone provokes different interpretations and hence different reactions depending on the context of the killing. Shooting someone dead in the street carries a different meaning from killing a soldier in wartime or killing a terminally ill loved-one. In Switzerland, for example, “assisted suicide” (helping someone take their own life) is not a criminal offence whereas it may be interpreted and classified as murder in the UK.

If behaviour itself (what someone does) doesn’t have the quality of deviance, it follows that this quality has to be found elsewhere, in terms of how behaviour is defined, or socially constructed, as deviant. This follows because, as Becker (1963) argues, ‘deviance is not a quality of the act the person commits, but rather a consequence of the application by others of rules and sanctions to an “offender”’. From this angle, explaining deviance does not involve looking at the individual and what they do (since, as we have seen, the same behaviour can be interpreted as both deviant and non-deviant); how others react to what someone does holds the key to understanding and explaining deviance.

As Durkheim (1895) explains, ‘What confers (criminal) character…is not some intrinsic quality of a given act but the definition which the collective conscience [society] lends it.’

While the ideas of relatively and social construction don’t necessarily prevent explanations for crime and deviance based on the characteristics (biological, psychological and sociological) shared by deviants, it does mean that any understanding of deviance must also include examining how and by whom rules are created and enforced since, as Becker (1963) argues, ‘Social groups create deviance by making the rules whose infraction constitutes deviance.’

Methods for Measuring Crime

Young (2001) suggests three main waysto calculate and quantify crime:

  • official statistics
  • victim surveys
  • self-report studies

Official crime statistics

Official crime statistics are published twice-yearly by the Home Office and cover a variety of categories (such as robbery, fraud, violent and sexual offences) that constitute officially recorded crimes (these are crimes reported to, or by, the police).


Crime statistics have practical advantages for researchers based around the fact that, as a secondary source funded, collected and published by government, a huge amount of time, effort, resources and expenditure is saved. They also avoid ethical problems, such as putting the researcher in danger, associated with some other methods of collecting data about crime and criminals.


Crime statistics also have methodological advantages built around their reliability. Simmons et al. (2003) note that the adoption of the National Crime Recording Standard (NCRS) was designed to standardise the ‘number of crime-related incidents reported by the public to the police and the number of crimes recorded by police’. In other words, it represented an attempt to ensure a standard system of recording across all police forces in England and Wales ‘to promote greater consistency between police forces in the recording of crime’.

While the NCRS has not completely solved the problem of reliably recording crime, it has gone some way towards ensuring ‘the integrity and consistency of police recorded crime figures’. In broad terms, data are collected in the same way from the same sources using the same definitions of crime (although definitions do sometimes change, these are clearly notified in the data). Kershaw et al. (2008) note how, ‘Trends in police recorded crime figures are affected by changes in police activity, coverage, public reporting and recording practices’ (the way in which police practices, such as ‘cuffing’ and ‘coughing’ affect statistical reliability are discussed in a later post).


Crime statistics are also broadly representative of the geographical distribution of crime across, for example, English counties, and provide a general indication of national crime levels (there are important qualifications we will consider in a moment). These features mean that crime statistics can be used for comparative purposes, to identify patterns and trends in criminal behaviour (such as different homicide rates) in both our society over time and between societies. As Kershaw et al. argue, ‘Police recorded crime statistics provide a good measure of trends in well-reported crimes, are an important indicator of police workload, and can be used for local crime pattern analysis.’

A further form of comparison involves the idea that local and national crime statistics can function as a benchmark against which to measure changing perceptions of crime. Green et al. (2010), for example, note that around 75% of tabloid newspaper readers believe crime is increasing, whereas crime statistics show it has fallen consistently over the past 15 years.


Validity is one area where crime statistics are frequently seen to be methodologically lacking: do they measure what they claim to measure? The broad answer is they do not (although there are significant exceptions noted below). While crime statistics provide valuable data about crimes reported to and recorded by the police, they tell us little or nothing about the dark figure of crime – crimes neither reported nor recorded. This is sometimes called the iceberg effect, in which reported and recorded crimes represent the ‘tip of the crime iceberg’ and the true extent of crime is effectively hidden from view.

The ‘dark figure’ of crime

Unsurprisingly, estimates of the ‘dark figure’ vary:

  • Young (2001), for example, suggests around 75% of all crime ‘is in the dark figure’
  • Kershaw et al. (2008) suggest the dark figure is around 60% (a figure based on analysis of British Crime Survey data)

However, they also note that for some individual crime categories validity is much higher:

  • homicide statistics are generally valid because the difficulties involved in disposing of a body make this a hard crime to conceal.
  • burglary statistics (around 75% reported) and car crime (around 95% reported) are also reasonably valid – people report these crimes to satisfy insurance requirements.

At best, therefore, crime statistics give us a reasonable measure of reported crime and while this is useful – Kershaw et al. note, ‘recorded crime statistics provide the only measure of homicide and also the only reliable measure of relatively rare crimes such as robbery’ – we need to be aware of their limitations. These are both technical and methodological:

  • Technical limitations: Simmons (2000) notes that although some police forces record‘every apparent criminal event that comes totheir attention’, the majority do not. An offence may be classified as ‘an incident’,which does not appear in the crime statistics.
  • Methodological limitations: crime statistics underestimate the extent of both minor crime (Kershaw et al. suggest that ‘vandalism, assault without injury and theft from the person’ are type of offending least likely to be reported) and some more serious forms of crime. These include some types of fraud, where people are either unaware a crime has been committed or, as Simmons suggests, businesses may want to avoid bad publicity from a police prosecution.

Reasons for under-reporting

There are a range of reasons for the under-reporting of crime. A victim may, for example, resolve the issue by confronting the offender. Alternatively, a victim may fear reprisal from the offender if they involve the police (something that applies to child abuse as well as more obvious forms of personal attack).

Sexual offences are, Simmons suggests, the least likely of all crimes to be reported. In terms of something like rape, a victim may decide not to prolong the memory or trauma of an attack; alternatively, they may believe the police will not treat them with consideration and sympathy.

With some forms of crime, such as street robberies, vandalism and the like, the victim may lack confidence in the ability of the police to catch the offender and unless an insurance claim is involved victims have little incentive to report such crimes.

Some crimes, such as prostitution and drug dealing, involve a ‘conspiracy of silence’ between those involved – someone buying illegal drugs is unlikely to report the offence, even in cases where they are robbed by a dealer. These types of crime are sometimes called ‘victimless’, mainly because the victimisation goes unreported.

Finally, for some types of crime, such as fraud, the victim may be ignorant of their victimisation.

Reasons for over-reporting

The over-reporting of crime is also a problematic area, one related to possible artefact effects – the idea that apparent increases in crime simply reflect changes in the way data are collected.

Simmons et al. (2003), for example, note the adoption of NCRS involved ‘a more victim-oriented approach to crime recording’, one based ‘more on the victim’s perception of a crime occurring rather than the police satisfying themselves that a crime had indeed taken place’. In other words, where someone defines themselves as a victim of crime the police must record their victimisation – a recording change that affects the comparability and validity of official crime statistics over time. Behaviour that prior to NCRS would not have been recorded as a crime, could now be recorded as such.

Similarly, Kershaw et al. note that proactive policing can result in an apparent increase in crime. By committing more resources to tackling specific crimes (such as burglary), the police discover ‘more crime’ by digging into the ‘dark figure’, which leads to ‘an increase in crimes recorded without any real change in underlying crime trends’.

Interpretation of data

Both over- and under-reporting illustrate the point that although quantitative data are normally considered more objective than qualitative data, the significance of any data must always be interpreted by the researcher – they have to decide what the data mean. A statistical rise in crime, for example, may result from:

  • a real rise
  • a different way of defining and counting crime
  • the police targeting certain types of crime

In addition, quantitative data tell us little or nothing about the reasons for people’s behaviour. While we can objectively quantify things, like the number of year-on-year murders, this tells us very little about why people kill each other.

Victim surveys

Victim surveys record crimes peoplehave experienced, but not necessarilyreported to the police. This is achieved, as with the government-funded British Crime Surveys (BCS), by interviewing people about their experience and perception of crime, as well as their attitude towards what Kershaw et al. (2008) term ‘crime-related topics (e.g. anti-social behaviour, the police, and criminal justice system)’.

The BCS is a major source of official data about crime victimisation that covers England and Wales (biannually1982–2000 and annuallysince) and involves interviewing nearly 50,000 people aged 16 or over (the survey was extended in 2008 to cover those under 16).

Alongside such surveys, local crime surveys focus on particular areas. The Islington Crime Surveys (Jones et al. 1986, 1990) and Policing the Streets (Young, 1994) areprobably the most well-known. Local surveys use similar techniques to their national counterparts, but on a smaller scale – Policing the Streets, for example, surveyed 1000 people in the Finsbury Park region of London.


Victim surveys have three important advantages:

  • They tell us something about levels of unreported crime (the ‘dark figure’).
  • They provide data about those at risk of victimisation and the measures they take to reduce their chances of victimisation.
  • They tell us about perceptions of crime (how much crime people believe takes place) and the fear of crime (the extent to which people believe they are likely to be a victim)

Spatial distribution and targeting

Victim surveys, in this respect, potentially tell us something about ‘real’ levels of crime and victimisation – something official crime statistics have, until very recently, told us little about. They also enable the pinpointing of spatial distributions of crime – the areas, for example, where crime is highest or where particular types of crime are most likely to take place.

This ability has started to influence police behaviour, particularly in relation to spatial targeting, such as the monitoring of known crime ‘hot-spots’, like particular areas of a town or city with lots of pubs and clubs. Using CCTV, plain-clothes police and so forth, the police are able to identify situations likely to quickly spiral out of control and, by rapid interventions, prevent more serious criminal behaviour.

Patterns and trends

Kershaw et al. suggest British Crime Surveys have specific methodological advantages: they provide broadly reliable measures of both victimisation and general crime patterns and trends, based on a broadly representative sample of adults aged 16 and over, with a response rate of around 75%.

Such surveys also have greater validity since they pick-up unreported and unrecorded crimes. As Hughes (2009) notes, victim surveys suggest the ‘real level’ of crime in our society is between two and three times greater than levels reported by official crime statistics.

However, as Kershaw et al. argue, the value of the British Crime Survey is not that it necessarily produces a more accurate total crime count, but rather that it provides ‘robust trends over time for the crime types it covers’. One reason for this is that the BCS is not affected by things like increased public perception and reporting of certain crimes, pro-active forms of policing or changes to the way crimes are recorded.


While victim surveys add to our knowledge and understanding of crime, they are not without their problems – one of the most significant, according to Hughes and Church (2010), being their coverage: ‘surveys are generally restricted to crimes against adults living in private households and do not include some types of crime’. Excluded crimes include:

  • commercial victimisation, such as thefts from businesses and shops or frauds
  • victimless crimes, such as drug possession
  • homicides
  • crimes by and against young people under 16 (before 2008)

This is significant for two reasons:

  • Victim surveys, by definition, bias the data towards crimes ‘known to the victim’, which exclude, Davies et al. (2003) note, important forms of crime: homicide, bank robbery, shoplifting and crimes against public sector property (arson, criminal damage, theft).
  • The picture we get of both victims and perpetrators is similarly uncertain: in terms of social class, for example, where crimes committed by the middle and upper classes, such as fraud, insider trading and corporate crime, are excluded from victim studies, the idea of crime as a ‘lower class phenomenon’ is reinforced – an idea that has significant consequences for some theories of crime.


Lea and Young (1993) also suggest that the validity of victim surveys, when compared to official crime statistics, is overstated – with the implication that the significance, if not necessarily the extent, of the ‘dark figure of crime’ is similarly overstated. While victim surveys dig deeper into ‘the “real” rates of crime between groups’ this is not particularly significant for three reasons:

  • It amplifies differences in criminality between different social groups: ‘The crime rate of old ladies is no doubt actually very low, but it probably appears even lower in the official statistics because of the police disinclination to suspect or arrest elderly persons. And as far as lower-working-class youths are concerned, the exact opposite is true.’
  • If crime statistics had greater validity, all that would happen is that the true level of middle/upper-class crime would be revealed – but so too would far higher levels of working-class crime.
  • As Davies (2003) argues, victim surveys only dig into the dark figure for some crimes. He estimates that between 11 and 33 million crimes each year are not picked up by victim surveys.


Three further methodological issues surrounding the validity of victim studies are highlighted by Mason (1997) in terms of memory, meanings and motives.

Memory: Asking people to remember events can involve

  • selective recall
  • false memory syndrome

When people feel under pressure to ‘remember’ victimisation they frequently do so in a selective way by reassessing events and reinterpreting them in a new light. Thus, an event not originally seen as criminal may come to be defined as such as a consequence of being interviewed.

Taylor (2012) points to a general criticism of this method when he argues that ‘even a nodding acquaintance with the psychology of perception and the work of Elizabeth Loftus on memory, for example, suggests that “telling it like it is” may not be quite that simple’.

Meanings: Young (1994a) argues that where the same behaviour (such as ‘being hit’) can have a different meaning for different people it creates a reliability problem. Working-class respondents, for example, are less likely than their middle-class counterparts to classify certain types of behaviour as ‘violent’ or ‘criminal’ because they have a greater tolerance of violent behaviour.

Ditton’s (1977) study of ‘pilfering’ and ‘fiddling’ also highlights class differences in behaviour, tolerance of criminality and how language is used to describe crime. He documented, for example, how ‘workplace fiddling’ is seen by those involved as part of the normal workplace experience – as a taxi-driver respondent forcefully expressed it when asked whether he ‘felt like a criminal’ because he worked illegally while unemployed and receiving State benefits, regularly overcharged customers and also defrauded his employer:

‘Don’t be f****** daft!’

‘No, but it’s breaking the law, isn’t it?… So why don’t you feel like a criminal?’

‘Nobody even thinks of it.’

Motives: People may be unwilling to talk about their victimisation. In the past, for example, British Crime Surveys underestimated the extent of domestic violence because the victim was understandably reluctant to admit their victimisation in the presence of the offender (their partner). Refinements in interviewing technique have gone some way to resolving this particular problem. This doesn’t mean that people’s motives for ‘acknowledging victimisation’ are not many, varied and difficult to quantify.

Self-report surveys

Self-report surveys are based around interviews or anonymous questionnaires and ask respondents to admit to crimes they have committed in a given time period. Such surveys provide us with data about the social characteristics, such as class and ethnic background, of offenders missing from other survey methods.


Thornberry and Krohn (2000) argue that self-report surveys are generally useful because they give access to data on a range of ‘sensitive topics’, such as offending. The ability to get both close to the ‘source of criminal behaviour’ (the actual offender) and collect data anonymously adds to their potential validity. Campbell (1981), for example, found far higher levels of female self-reported crime than was revealed by official crime statistics – something indicative of how self-report studies can give an insight into the dark figure of crime.

These surveys can also provide information about the frequency and seriousness of different forms of offending. Wikstrom’s (2012) ten-year longitudinal study, designed to investigate ‘key individual and environmental factors which influence social behaviour during the transition from child to adulthood’, used self-report behavioural questionnaires to gather comparative data about levels of offending among the young males in the study.


Young (1994) argues that the reliability and validity of self-report surveys can be criticised in the context of their representativeness. The majority of surveys focus on the behaviour of young people and although this tells us something about their offending (in terms of class, gender and ethnicity, for instance), it’s difficult to see how findings can be reliably generalised. Weitekamp (1989) argues that while surveys discover a mass of relatively trivial delinquent behaviour, they miss a vast range of offending associated with adults.

Access problems mean that although these surveys tell us something about the ‘crimes of the powerless’ – particularly working-class youth – they tell us little or nothing about the ‘crimes of the powerful’; the latter rarely, if ever, participate in this kind of research, something that casts further doubt on its representativeness.

Finally, self-report surveys suffer from various types of validity problem:

  • Jurgen-Tas et al. (1994) note strong evidence that ‘prior contacts with the juvenile justice system’ make offenders less likely to participate in self-report surveys, which once again contributes to their unrepresentativeness.
  • Young argues that the setting of many studies (‘a middle-class interviewer, often in the official setting of the school’) creates ‘an optimum socially structured situation for fabrication’ – respondents are consciously or unconsciously encouraged to lie.
  • Jupp (1989) suggests that respondents tend to admit honestly to trivial offences but are less willing to admit to serious offences.

Next Notes

Crime Trends and Patterns.

Stay Updated

Enter your email to be notified when we post something new:

Archived Posts