CRIME DATA
There are three types of lies. Lies, Damn lies, and Statistics (Mark Twain)

    No way of measuring crime is perfect, and it is extremely difficult, if not impossible, to know exactly how much crime is going on in any jurisdiction at any time.  First of all, crime is a label slapped on some behaviors and not on others.  There are known inaccuracies in the labeling process.  Secondly, much crime goes undetected.  Thirdly, some crimes are not reported to police.  Crimes that go undetected and unreported obviously cannot be counted.  Finally, the police themselves may, for various reasons, not record something as a crime, or inaccurately report something as a crime when it is not.  Criminologists refer to all crime that escapes counting for any of the above reasons as the dark figure of crime.  Do we really know how much crime is out there?  Is what we really know only the tip of the iceberg?

    Here's an example of a typical police department's manipulation of crime reporting.  In 2004, Atlanta's own police chief ordered an audit be carried out for crime reporting from previous years.  The audit found that the city police department had underreported crime for many years in an attempt to boost the city's appeal to tourists and win the site selection process for the 1996 Olympics.  The audit was carried out by the New York-based auditing firm Linder & Associates.  It found that in many cases (22,000), crime reports were simply lost because the department was overworked, understaffed, and sloppy.  However, in many other cases, crime incidents were deliberately downgraded, underreported, or discarded in order to make the city look good on paper.  The result of all this deliberate activity was that the crime rate in Atlanta was reported as 7% lower than it should have been.  The standard FBI crime report margin of error is 2%.  It may also be noted that Atlanta has about 15 law enforcement agencies (university, hospital, public transportation police, etc.) which do not send their crime reports into either the Atlanta police or the FBI.  

POLICY IMPLICATIONS

    Accurate crime data serve important purposes.  It is used by public agencies and officials in determining policies, budgets, legislation, funding priorities, and evaluation of existing programs.  If the government was not in the business of trying to collect accurate crime data, it would have to rely upon mostly inaccurate and  unrealistic public opinion polls.  The agencies involved run all the way down from the White House Briefing Room to each state's Statistical Analysis Center.  It may or may not also be apparent that crime data is often used to convince voters of the success or failure of crime control policies.  Private, nonprofit agencies also use crime data to justify their position on crime issues. The web has an abundance of victim-oriented sites in this regard.

    Social science researchers also analyze crime data in comparison to other data, like sociologists interested in "social indicators."  A sociologist, for example, might compare emergency hospital room admissions for drug related problems with the numbers for drug related arrests by police. A criminologist might stick to criminal justice statistics and look at rising rates of the incarcerated offender population and how they relate to crime rates overall.

MAIN SOURCES OF CRIME DATA

    NCJRS, the National Criminal Justice Reference Service, maintains one of the largest repositories of web sites devoted to criminal justice statistics.  NCJRS is probably the most extensive source of information on criminal justice in the world. It's a group of clearinghouses supporting all the bureaus of the U.S. Department of Justice, Office of Justice Programs, the National Institute of Justice (NIJ), the Office of Juvenile Justice and Delinquency Prevention (OJJDP), the Bureau of Justice Statistics (BJS), the Bureau of Justice Assistance (BJA), the Office of Victims of Crime (OVC), and the Office of National Drug Control Policy (ONDCP).

    Academics and more serious researchers often make use the Sourcebook of Criminal Justice Statistics, which highlights and tracks some of the most important tables or indicators in the field.  Students often report that the Sourcebook is an excellent place to download figures and graphics for term papers.

    To go right to the source of crime data collection, however, one uses the FBI's Uniform Crime Reports (UCR) or BJS's National Crime Victimization Survey (NCVS).  It is customary for students to learn and remember various details about the UCR and NCVS as they are the oldest and most well-known methods of crime data gathering.

THE UNIFORM CRIME REPORTS

    The UCR program was started in 1929 by order of the U.S. Attorney General, and the FBI is responsible for administering the program.  Over 17,000 law enforcement agencies (city, country, state) report to the FBI monthly the number of crimes known to them in their respective jurisdictions.  These crimes are reported as rates (which is the number of crimes per some unit of population, such as 1,000 people for a small jurisdiction, and 100,000 people for a large, or national jurisdiction).  The idea since the beginning is that the UCR program would enable comparisons and serve as an indicator, estimate, or index of the nation's well-being or moral health, at least when it comes to crime.  Eight (8) offenses -- murder, rape, robbery, aggravated assault, burglary, larceny-theft, motor vehicle theft, and arson -- became the basis of this crime index, also known as Part I offenses.

    UCR terminology includes things such as the hierarchy rule, which means that when multiple crimes are known to have been committed by an offender the same year, only the most serious crime is counted.  The hierarchy rule is considered a limitation of the UCR.  Another limitation is the limited amount of data on arrestees and victims (except in homicides), and the complete absence of any data on victim-offender relationships.  Nonetheless, the UCR has been helpful in comparing the number of arrests with the number of crimes known to police.  That comparison is the traditional measure of police effectiveness, and is known as the clearance rate.  There are cases where the police know who did it, but cannot make an arrest for one reason or another (suspect fled the country, committed suicide, died, or was convicted in another jurisdiction), and these situations are called exceptional clearances.  Below is a ranking of the (most frequently occurring) Index (and the less serious Part II) offenses, along with clearance rates for the year 2000:
 

    UCR data provide information on patterns and trends in various offense categories, as well as patterns in the overall crime trend.  Murder, for example, is known to be more common in southern states (a fact that theorists of the "Southerness hypothesis" have been trying to explain).  We also know murder is a crime committed by friends or acquaintances; few strangers are involved.  July and August produce the highest number of murders.  Rape is a crime where it is estimated only one out of four offenses are reported, so use of UCR data on rape might not be wise, but according to it, rape follows some of the same patterns as murder.  Robbery occurs quite frequently, primarily in urban areas among the young, and 20% of the time, the victim is shot, and 3% of the time, the victim is raped.  Aggravated assaults ("aggravated" because it involves a weapon or results in hospitalization) are fairly easy to solve, and you'll notice the clearance rate is almost as high as murder.  Burglary, believe it or not, tends to happen more frequently during daytime than nighttime, according to UCR records.  Larceny-theft (larceny being the taking of anything not belonging to you and theft being the dollar amount) is the most frequently reported crime, and motor vehicle parts or contents tend to be the most common item taken.  Motor vehicle theft tends to be frequently reported but rarely cleared.

    The overall crime trend may or may not be a useful figure for indexing moral health, but it does tend to show certain regularities that may be of other theoretical interest, or more practically, for purposes of agency budgeting, personnel deployment, and tourism planning.  Crime overall (both violent and property, but mostly violent) tends to exhibit a strong seasonality factor.  For example, summertime is definitely crime time, and the crime season usually starts on Memorial Day weekend.  The regularity in the cycle goes like this -- crime rates bottom out during January and February; in March they start to increase, increasing regularly over April and May; with June being the month when the crime rate usually rises above the national average; and with August being the peak month.  In September, the crime rate usually starts to decline.  Nobody is really sure how to explain this regularity, and explanations can be found ranging from thermal stress to the effects of air pollution, although opportunity may be the only important variable.  

NEW METHODS OF CRIME REPORTING

    NIBRS stands for National Incident-Based Reporting System and is a new FBI method of collecting data that was introduced back during 1985 and slowly became implemented in the states.  About 4,300 law enforcement agencies (city, county, state) contribute NIBRS data to the FBI every year.  This new system eliminates the hierarchy rule and collects a significant amount of more information about the crime, such as victim-offender relationship, property details, and other details about the offender such as motivation.  Known as IBR (Incident-Based Reporting), this strategy allows analysis of the incidence of crime (# of events) rather than simply calculating the prevalence of crime (# of people).  Incidents are the number of cases known to police, and incidence is treated as relationship linkages.  For example, if five offenders attack five victims, the UCR would only count five crimes, but NIBRS would look at the age, race, sex, and ethnicity (in NC) of each offender, plus a relationship linkage between each victim and offender, and the database might be able to come up with a count of as many as ten or twenty crimes based on relationship linkages between these elements.  Clearance rates are still based on arrest, but NIBRS "Group A" offenses consist not only of arrests, but investigative reports and incident reports as well, which include attempts and suspected events.  "Group B" offenses are based solely on arrest.  This method is ideal for strategic and tactical crime analysis at the local and regional level, and the database is robust enough to allow creation of new categories, such as Crimes Against Society, or for the researcher to explore other interesting relationships among the variables, usually online via the NACJD website.  The crimes that make up Group A and B offenses are quite interesting, and of modern concern, such as pornography, counterfeiting, prostitution, embezzlement, bad checks, weapons violations, and Peeping Tom activity.  The following is a table of Group A and Group B offenses, and items marked with an asterisk generally do NOT have investigative or incident reports associated with them, although sometimes they do:
 

    In 1990, Congress also passed two new laws, the Campus Crime and Security Act, which requires colleges to issue annual reports, and the Hate Crime Statistics Act, which requires police to document hate-related incidents. During the year 2000 on college campuses, there were 400 murders, 4,000 rapes, 13,000 robberies, 19,000 aggravated assaults, 31,000 motor vehicle thefts, and 70,000 burglaries.  With regard to hate crime, police documented 8,150 incidents, including 19 murders, a few robberies and rapes, but the vast majority were assaults.  The most common motive was racial hatred (60% of the time), religious bias (17% of the time), and based on sexual orientation (homophobia, 16% of the time).  The remainder of cases had undeterminable motive.

THE NATIONAL CRIME VICTIMIZATION SURVEY

    The NCVS began operation in 1972, and continues to surprise analysts and researchers with its results.  Cities and towns that the UCR showed to have little crime turned out, via the NCVS, to actually have high rates of crime.  As a general rule for most places, NCVS data will most usually show twice as much crime as the UCR indicates.  NCVS interviewers consist of both Justice Department employees and U.S. Census Bureau employees who interview 50,000 households twice each year.  Households are rotated randomly every three years.  The BJS publishes the report, entitled Criminal Victimization, every year.  The NCVS uses the hierarchy rule and records the same Part I offenses as the UCR, but it doesn't ask about offenses that might be too sensitive in an interview situation, like murder and arson.  Not only does it not measure homicide, but it doesn't measure commercial crime (such as burglaries of stores), hence the name of NCVS reports are called Households Touched by Crime.  It does a good job of focusing on precisely the amount of value attached to any property stolen.  The NCVS collects information on crimes suffered by individuals and households, whether or not those crimes were reported to law enforcement.  It estimates the proportion of each crime type reported to law enforcement, and it summarizes the reasons that victims give for reporting or not reporting.  It provides information about victims (age, sex, race, ethnicity, marital status, income, and educational level), offenders (sex, race, approximate age, and victim-offender relationship), and the crimes (time and place of occurrence, use of weapons, nature of injury, and economic consequences). Questions also cover the experiences of victims with the criminal justice system, self-protective measures used by victims, and possible substance abuse by offenders. Supplements are added periodically to obtain detailed information on topics like school crime.

    The patterns and trends from the NCVS indicate that 25% of all households are touched by crime every year.  That's a total of 25 million victimizations, far greater than totals from the UCR.  Below is a table indicating the most frequent crimes, according to NCVS data, 2000:
 

    As indicated, the amount of crime indicated by the NCVS is significantly greater than that indicated by the UCR.  The only exception to this is motor vehicle theft, which can probably be explained by insurance companies requiring a police report, which citizens dutifully do, even to discover later their vehicle was not missing.  The NCVS relies upon random sampling of victims and their memories.  Some respondents may also lie, cheat, or exaggerate (called falsification), but the main memory problems are decay (incident happened recently, but person forgot it) and telescoping (incident happened long ago, but person remembers it like yesterday). 

SELF-REPORT CRIME SURVEYS

    A third method of collecting crime data is to make up your own questionnaire or survey.  This is exactly what is done by most academic researchers and private foundations, and a few government agencies have been doing it since the 1970s.  The Monitoring the Future project and National Youth Survey are the best known examples.  A few private projects have been federally funded, and several researchers hope to devise the perfect measuring instrument.  Much of the information used in recidivism studies is derived from self-report data.  A self-report survey is typically administered to high school students, although sometimes college samples are involved.  Other samples have involved incarcerated prisoners as well as attempts to find uncaught criminals.  Regardless of the sample population, the purpose is to get respondents to read the questions and mark down how many times they have committed a serious offense and gotten away with it.

    The results of most self-report studies are shocking.  They typically indicate, that for any population (even a law-abiding one), about 90% of the people in the sample have committed a crime for which the punishment is more than a year in prison.  This has led many people to question the accuracy of such surveys.  The main problems are, of course, validity and reliability, but it may be just as true that people lie or exaggerate on these questionnaires (falsification).  There are few self-report studies of white-collar crime, and it is easy for researcher bias and measurement error to creep in.  Golub et. al (2002) probably provided one of the best summaries of the limitations of self report data when they found, compared to criminal records checks and urinalysis screens, things like marijuana use were most often accurately disclosed while involvement in violent crime was least often disclosed, but if prior conviction matched self-reported data on prior convictions, then there was reason to believe that the rest of the data was fairly reliable.

OTHER TERMS IN THE STUDY OF CRIME DATA

    There are a number of terms one will encounter with analyses of crime data, and a variety of definitions abound.  Here are some of the most common terms with their definitions.  The frequency of crime, or frequencies, refer to the time span the researcher, usually a self-report researcher, is interested in.  The most common time spans are "ever" and "last year", but there is also one called "last month", with last year taken as a measure of seriousness and last month interpreted as a measure of persistence.  Hence, one talks about a frequent, serious, or persistent offender as one who scores high on those survey items.  Most surveys are retrospective; that is, they ask the respondent to report crime they did in the past.  There are also surveys which are prospective, which ask the respondent how much crime they think they'll commit in the future.  Prospective surveys are interesting, especially when the researcher combines them with items measuring possible causes of crime.

    Criminologists often distinguish between prevalence and incidence, which is the distinction between the number of people committing crime (known offenders) and the number of offenses committed by criminals (known offenses).  These terms are also known as participation and lambda, respectively.  Dividing the number of known offenses by known offenders and multiplying by a constant, such as 100,000 is a way of calculating crime rates (incidence divided by prevalence times 100,000).  Lambda is the more important term for those interested in criminal career research, for those interested in identifying high-rate offenders whose criminality does not seem to decline with age.  In most studies, lambda is usually calculated as between 3-8 offenses, which means that for every offender apprehended on the instant offense, there are an additional 3-8 other crimes they are responsible for in the past year.  Deterrence can be measured by setting the arrest rate equal to the product of lambda and the probability of apprehension (which generally runs at 1:1000 for most offenses).  Incapacitation can be measured by estimating career length.

    Career criminal research involves some other terms one might encounter.  The notion of chronicity (as in chronic offender) is commonly seen in the context of 7-70 theory (Wolfgang et. al. 1972) like "seven percent of offenders commit 70% of the crime", which is well-established in the literature.  The notion of velocity (as in the velocity of crime) refers to calculating the average amount of street time to the next offense.  Hence, a chronic offender, in and out of jail quite often, who averages 7-8 days between offenses, is less of a high-velocity offender than one who averages 2-3 days between offenses.

    Chronic, high-velocity offenders out on the street are probably a rarity, however.  Most research of this type is done on incarcerated offenders, which most likely produces overestimates.  On the street, the study of what keeps people from a life of crime, or turns them away from it is called the desistence field of research.  Theoretically, such variables as vulnerability and insulation (also called buffering) are important, which refer to conditions or states that turn good people bad or bad people good.  However, researchers have been more interested in the deterrent effect of imprisonment, which has been estimated as having no more than a 10-20% deterrent effect per year (Shinner & Shinner 1975).  The fact of the matter is that more people age out of crime than for any other method of desistence (Gottfredson & Hirschi 1990), and the so-called age-curve constant is remarkably stable, even for high-rate, persistent offenders.  In fact, as Golub's (1990) research shows, imprisonment may actually disrupt the normal pattern of desistence.       

INTERNET RESOURCES
Bureau of Justice Statistics Home Page
Campus Crime and Security Website
Crime Data MegaSite
Crime Statistics Guide
FBI Uniform Crime Reports
Forecasting of Crime Seasonality (pdf)
National Archive of Criminal Justice Data
Sourcebook of Criminal Justice Statistics
When Incarceration Increases Crime

PRINTED RESOURCES
Bachman, J. et. al. (2000). Monitoring the future. Ann Arbor, MI: Inst. for Social Research.
Blumstein, A., J. Cohen and D. Farrington (1988). Criminal career research. Criminology 26:1-35.
Cheatwood, D. (1988). "Is there a season for homicide?" Criminology 26(2): 287-306.
Elliott, D. (1994). "Serious violent offenders." Criminology 32:1-21.
Federal Bureau of Investigation (yearly). Crime in the United States. Washington D.C.: GPO.
Golub, A. (1990). The termination rate of adult criminal careers. Pittsburgh: Carnegie Mellon.
Golub, A., Johnson, B., Taylor, A. & Liberty, H. (2002). "The validity of arrestee's self-reports." Justice Quarterly 19(3): 477-502.
Gottfredson, M. and Hirschi, T. (1986). "The true value of lambda would appear to be zero." Criminology, 24:213-234.
Gottfredson, M., & Hirschi, T. (1990). A general theory of crime. Stanford, CA: Stanford University Press.
Maguire, K. & A. Pastore (eds.) (yearly). Sourcebook of criminal justice statistics. Washington D.C.: GPO.
Mosher, C., Miethe, T., & Phillips, D. (2002). The mismeasure of crime. Thousand Oaks, CA: Sage.
Shinnar, S., & Shinnar, R. (1975). "The effects of the criminal justice system on the control of crime: A quantitative approach." Law and Society Review, 9, 81-611.
U.S. Dept. of Justice (yearly). Criminal victimization in the United States. Washington D.C.: GPO.
Wolfgang, M., Figlio, R., & Sellin, T. (1972). Delinquency in a birth cohort. Chicago: University of Chicago Press.

Last updated: July 02, 2007
Not an official webpage of APSU, copyright restrictions apply, see Megalinks in Criminal Justice
O'Connor, T.  (Date of Last Update at bottom of page). In Part of web cited (Windows name for file at top of browser), MegaLinks in Criminal Justice. Retrieved from http://www.apsu.edu/oconnort/rest of URL accessed on today's date.