Crime Statistics Research Paper

The collection and monitoring of official crime statistics has largely been taken over by government and government sponsored agencies such as the US Bureau of Justice Statistics or the UK Home Office (Coleman and Moynihan 1996). These agencies draw on the major criminal justice institutions such as the police, prisons, and the courts for these official statistics. On the surface the collection of such data might appear straightforward and relatively uncontentious yet in reality it remains one of the most controversial areas in the analysis of crime. Official crime statistics have been subject to critique since the mid-nineteenth century through to the present, the beginning of the twenty-first century (Coleman and Moynihan 1996).

There are a variety of factors that contribute toward this controversy (Kempf 1990, Jackson 1990). Some of the more important reasons include:

(a) The study of crime itself is often controversial.

(b) What is defined as a ‘crime’ is a social construction and not an empirical fact.

(c) There is not always consensus on how to collect crime statistics.

(d) There is no one data source that can address all the issues raised in the study of crime.

2. Official Crime Statistics

Administrative databases are used to generate official crime statistics yet they are first and foremost a management tool. As a result, using data generated from this system as the sole source of crime statistics is problematic. Unlike many other economic and social activities, the act of committing a crime can result in the loss of liberty and in some cases the loss of life. Understandably, most people do not admit to committing crime. As a consequence detection of crime is dependent on victims reporting such activities to the police, and the police successfully apprehending the suspects. Police make decisions about whether to proceed with charges, then prosecutors make decisions about whether to continue with the charge and if so which charges they will pursue. Finally, the courts then make a determination as to guilt or innocence. There are many factors that intervene at different points in this process ultimately affecting the ‘numbers’ that appear in official statistical collections. Because of the potential for error at various stages through the recording process official homicide statistics are largely regarded as having the most credibility.

3. Police Data

Counting rules used by the collection agencies complicates interpretation of police statistics. Data can be confusing to the novice as it can on occasion refer to the offences/incidents and at other times to the offenders and victims. For example, one offender may be responsible for 50 offences. Similarly, five people may be involved in one offence (e.g., a bank robbery by a group of people). Where an individual has committed a range of offences often only the most serious offence is recorded for statistical purposes. For example, prison statistics in Australia are based on the most serious offence of the prisoner. Thus if an individual is convicted on a range of offences on a particular day only the most serious offence is recorded. As a result official statistics can under-report the extent of crime in the community. Police data are limited by:

(a) Definitions of what constitutes a crime.

(b) Variability in reporting by citizens.

(c) Law enforcement organizational structures.

(d) Variations in counting rules across jurisdictions.

(e) Low rates of detection and clearance.

4. Court And Corrections Data

These particular administrative databases represent a subset of police data for they record crimes that have been processed further down the criminal justice chain. For example only one in 20 people who appear before the criminal courts in Australia serve time in prison. Data from these systems are used to produce crime statistics on matters relating to sentencing, prison composition, recidivism, probation, and parole. However, as with police databases they are highly restrictive in terms of the limited nature of the data recorded. Usually only the most basic demographic and arrest characteristics are collected. Counting rules also vary across the courts. In Australia, for example, the unit of count is the charges or appearances in the lower court while in the higher court it is distinct persons. Crime statistics from these databases will be biased as they are based on those who represent the most serious offenders and offences.

5. Improvements In Official Statistics

Despite these limitations there have been significant advances in the recording of crime within administrative systems. Many countries have moved or are moving toward collating national data on crime that is comparable across areas by standardizing definitions and counting rules (UNAFEI and AIC 1990). In addition jurisdictions are seeking to link data across criminal justice institutions so that a more complete picture of crime, offending, and offenders can be determined. Improved and cheaper technologies, and the development of statistical packages, have meant that analysts and researchers are demanding unit record files (URF) rather than aggregated statistics. As a result new administrative databases are being established to provide greater detail in response to research and policy demands. A major development has been the National Incident-Based Reporting System by the United States Federal Bureau of Investigation (Maxfield 1999).

There are three major factors that are driving improvements in crime statistics. The first is the demand for accountability by governments for the way in which scarce public money is spent. To justify their activities criminal justice institutions are increasingly relying on statistics to support their assertions and resource allocations. This results in both a greater awareness of statistics and a greater scrutiny of these same statistics. The second major factor has been the critical scrutiny of these data particularly by independent academic researchers. The third has been the development of technology that enables the cost-effective storage and retrieval of data.

6. Crime Victim Surveys

Researchers have long been concerned about a ‘hidden figure’ of crime that represents a large proportion of unrecorded crimes and criminals. Crime victims’ surveys from a range of countries have shown that the majority of people do not, for a variety of reasons, report crimes to the police. There are other biases in the official statistics. White-collar crime is rarely recorded or dealt with under criminal law (Sutherland 1949). Some researchers argue that the important feature of crime data is the long-term trend in crime and not the overall amount of crime.

Others have argued that even the long-term trends are difficult to interpret—is crime really increasing or is it a reflection of other things? For example, changes in police practices and priorities, or changes in the interpretation and classification of particular offences can help to explain changes in crime statistics. There have also been changes in the police management/processing of some forms of crime, such as rape and domestic violence, that have impacted on the number of recorded crimes.

Given the sustained criticism of official statistics governments and academics have sought ways to improve upon their collections. To address the issue of under-reporting of crime to police, crime victim surveys are routinely undertaken in many countries. Individuals within households are randomly selected and asked about their experience of crime and victimization. From this source of information the nature and extent of crime has been shown to consistently be more prevalent than appears from ‘official’ statistics. However, such surveys have also been criticized on a number of grounds including (Koffman 1996, Coleman and Moynihan 1996, Skogan 1981):

(a) The definition of what constitutes a ‘crime’ in the survey instrument does not always match the legal definition in the official statistics making comparisons difficult.

(b) Particular subgroups are either more or less likely to self-report particular crimes affecting the reliability of the estimates. For example, those with higher levels of education are more likely to report assault, while women are less likely to report rape domestic violence.

(c) Memory failure is a problem with respondents being more likely to recall incidents where the offender was not known to them than where the offender was known or related. There are also potential problems in terms of ignorance of events, deception, and ‘forward and backward telescoping.’

(d) Methodological problems include who should be interviewed in the household and the phrasing and order of questions.

7. Self-Report Surveys

In addition to victimization surveys there has been increasing use of self-report data from offenders. Dissatisfaction with official crime statistics has been a major factor in the development of offender surveys. This source of information has tended to be dominated by academics rather than government statistical collection agencies. One of the reasons is because these data need to be more carefully informed by theoretical models that can only come from an in-depth understanding of the substantive material. Self-reports are now regarded as the most common methodology for collecting data on crime statistics and large scale self-reported studies have taken place in a range of countries (see Junger-Tas and Marshall 1999).

Self-report data have generally confirmed that official data underestimates the level of crime in the community as described in official statistics. This particularly applies to victimless crime such as prostitution and illegal drug use. It is in the area of deviant behavior that self-reported crime statistics are superior to official statistics and these behaviors, if defined as illegal, are rarely reported to the police. Hence, self-report data are widely used to provide more accurate statistics on the prevalence and incidence of engagement in criminal activities by offenders. In addition self-report data have been used to:

(a) examine the correlates of offending;

(b) test theoretical models of criminal behavior; and

(c) evaluate intervention programs (Junger-Tas and Marshall 1999).

However, biases also apply to many self-report studies. Unless extraordinarily large sample sizes are obtained in population surveys the capacity to examine the correlates of offending or test theoretical models is limited by the highly skewed distribution of criminal behavior; this form of deviance is a rare event. For example, less than half a percent of the general population self-report having tried heroin in the past 12 months in household surveys in the USA and Australia.

To generate 100 people who have used heroin in the past 12 months would require a sample of 20,000 people. In addition most household surveys tend to under-represent those engaged in criminal activity; this is partly due to the sampling frame and partly due to the differential participation amongst particular groups (Weis 1986).

One solution to overcoming the problem of large sampling error is to use disproportionate stratified samples or to focus on specific groups such as incarcerated offenders. The latter will ensure that the sample has the desired characteristics but generalizability is severely limited.

Given the different methodologies employed to obtain crime statistics it is not surprising that the absolute rates vary between official and self-report data from offenders and victims. Consequently, researchers have advocated that relative rates should be used in making comparisons (see Junger-Tas and Marshall 1999). When this is done there is considerable consistency between the statistical sources from a range of difference countries (Junger-Tas et al. 1994).

8. Impediments To The Improvement Of Crime Statistics

There are three major impediments to the improved development of crime statistics:

(a) the application of ethics privacy informed consent provisions;

(b) access to primary or unit record data; and

(c) the rising costs of undertaking major research studies.

The movement toward increased ethics (or human) oversight of projects can be problematic particularly in the area of crime. In practice the self-report methodology asks people to incriminate themselves. This poses real dilemmas for ethics committees, many of who are not trained in the discipline. As a result committees may be reluctant to give ethics approval. It also poses problems for researchers who in many countries and jurisdictions do not have protection under the law; they may be legally required to release information that was gained under assurances of confidentiality to the informant. Instigating informed consent forms, particularly when it requires a sig- nature may conflict with assurances of confidentiality or anonymity thus affecting response rates. JungerTas and Marshall (1999) allude to the problem of informed consent on participation rates for delinquency studies. These issues militate against the study of crime and could well reduce the empirical study of crime and the collection of crime data outside of the official statistics.

The major improvements to crime statistics have largely occurred because of academic scrutiny of published statistics and secondary analysis of unit record data. The USA has had a long tradition of sharing and releasing statistical data for secondary analysis; this is not the norm. Access to official data collections for secondary analysis is often difficult and highly controlled by state institutions in many countries, like Australia. Until state institutions release full unit record data for secondary analysis this situation will continue to impede the improvement of national crime statistics.

Finally, funding constraints balanced alongside the increasing costs of data collection and research will affect both the government and nongovernment sectors in their ability to improve on their statistical collections. This is compounded by improvements in statistical methods that allow social scientists to undertake more sophisticated analysis. But this entails more sophisticated statistical collections and these are usually much more expensive.

Bibliography:

1. Coleman C, Moynihan J 1996 Understanding Crime Data. Open University Press, Buckingham, UK
2. Jackson P 1990 Sources of data. In: Kempf K L (ed.) Measurement Issues in Criminology. Springer-Verlag, New York
3. Junger-Tas J, Marshall I H 1999 The self-report methodology in crime research. In: Tonry M (ed.) Crime and Justice, A Review of Research. University of Chicago Press, Chicago, Vol. 25, pp. 291–367
4. Junger-Tas J, Terlouw G-J, Klein M W (eds.) 1994 Delinquent Behaviour Among Young People in the Western World. Kugler, New York
5. Kempf K L (ed.) 1990 Measurement Issues in Criminology. Springer-Verlag, New York
6. Koffman L 1996 Crime Surveys and Victims of Crime. University of Wales Press, Cardiff, UK
7. Maxfield M G 1999 The national incident-based reporting system: research and policy applications. Journal of Quantitati e Criminology 15: 119–49
8. Skogan W G 1981 Issues in the Management of Victimization. US Department of Justice: Bureau of Justice Statistics
9. Sutherland E H 1949 White Collar Crime. Dryden, New York
10. UNAFEI and AIC 1990 Crime and Justice in Asia and the Pacific. Australian Institute of Criminology, Canberra, Australia
11. Weis J 1986 Methodological issues in criminal career research, Vol. 1. In: Blumstein A, Cohen J, Roth J A, Visher C (eds.) Criminal Careers and ‘Career Criminals.’ National Academy Press, Washington, DC