Printer Friendly

Stalking security statistics.

THE USE OF PUBLIC SAFETY organizations' criminal event records is becoming a necessary tool in companies' risk-management programs. Much like market surveys, crime statistics can be used effectively in commercial site selection, insurance and premium setting, and other evaluations traditionally associated with capital investments unrelated to routine protection needs. There was a time, for example, when a shopping center developer only had to secure a promising location and commission a market survey and traffic count in deciding on the site for a new mall. Now, a developer probably should hire a security consultant to analyze whether a prospective site could be designated as either a high-crime area or a low-crime area.

A security consultant--whether in-house or outside--is called on to do one or more of the following: assess what types of threats or risks affect the assets to be protected, render an opinion on the probability of those threats or risks, and recommend a security or loss prevention plan to reduce the probability of those threats or risks. While not all security risks are crime oriented (such as natural disasters or fires), for the most part security risks entail crimes against people or property.

Today premises liability considerations are almost a requirement in business. Crime must be evaluated in preparing a security program that meets the threat of existing levels of risk and deters foreseeable adverse criminal incidents. The mere possession of crime statistics, however, is not sufficient. Without the capacity to establish patterns of crime properly and evaluate the impact of past and future crimes on a commercial premises and its invitees, statistics by themselves offer no insights and may even be misleading.

The professional law enforcement community generates a variety of crime data: offense categories, occurrence times, offender and victim characteristics, clustering, frequency, and clearance rates. Police calls are sorted by several activities, such as traffic, nuisance, domestic, crime, order maintenance, medical emergencies, and assorted administrative and judicial assignments. Bigger law enforcement agencies use computer programs and information management procedures to pull out various tracking formats to help in personnel deployment and routine administrative decision making.

Crime statistics are available from a number of sources, the chief repository being the Uniform Crime Reporting (UCR) program managed by the FBI for the nation's law enforcement and criminal justice establishments. Crime in America is published each year and covers the various crime and police data for most cities and suburban communities. Urban police departments routinely offer aggregate crime statistics for Part I offenses (the most serious offenses) and Part II offenses (the least serious offenses).

While these figures are helpful in making comparisons between cities and states, little data is actually useful for determining whether a particular commercial site is more or less susceptible to crime. Proper analysis of crime statistics for security decisions involves at least some geographic mapping, comparative measures, and examination of percentages of change over time and of frequency distributions by per capita ratios. Unfortunately, this information is not found in the annual reports of police departments. The information has to be developed from dispatch records, which are then coded by completed offenses according to UCR subcategories and the disposition classification given by field officers responding to the initial call.

Law enforcement agencies normally release UCRs every fall and provide the previous year's data for the media to compare the annual changes. Some elective agency heads withhold crime statistics to reduce political consequences or allay the public's fear of crime. Others exclude some categories and deny any public access to the data. In most cases, however, computerized records of generic crime statistics are available to security researchers and public safety professionals. That is particularly true in jurisdictions with emergency 911 service.

Complaints are usually organized by time, location, and category, including an item number for obtaining field incident reports; site-specific comparisons can then begin. Often, the security consultant must obtain original police incident reports for specific crimes to gather data about victim characteristics or extenuating circumstances. Security consultants must familiarize themselves with how crimes statistics are recorded in particular jurisdictions. In many instances, a crime address indicated on a printout will refer to where the crime was reported rather than where it occurred. Knowledge of police zones, districts, or beats is also invaluable in crime analysis.

Gaining access to crime information can be a tricky problem for private security practitioners. The courts hold management responsible for checking the criminal history of employees, but to protect the privacy of victims and arrestees, statutes often forbid law enforcement agencies to release such information. In addition, damage awards have been levied in inadequate security judgments against businesses that did not avail themselves of police statistics that might have defined a certain site as a dangerous or high-crime area, requiring enhanced protection for employees and guests.

While official crime statistics come almost exclusively from criminal justice agencies, other types of official data do exist. For example, certain crimes committed on college and university campuses may be handled internally and not reported to the local police. Though the practice is legally questionable, university administrators are often loath to report crimes to official law enforcement agencies because of the unfavorable publicity.

To legally avoid reporting, many universities are changing their security departments to police departments directly commissioned by the state as separate criminal justice agencies. It may be difficult in these cases for a security consultant to obtain accurate crime data. Although actual crimes (such as rape) may occur on campus, they may be classified as student conflicts and handled internally by the university. Unless a student or his or her parents report the crime to the police, it may never see the light of day. Internal crime reports usually have to be subpoenaed, and some other corroborative data (such as interviews with students) is often necessary to establish crime trends.

The security consultant must, of course, fit the data to the particular assignment at hand. In many instances official data on, for example, prior crimes by type reported at a site may be sufficient for an initial risk analysis. Here the consultant may be trying to get a feel for the types of crimes (armed robbery, commercial burglary, vandalism, etc.) that have occurred most frequently at the facility in question.

More detailed data may or may not be required later depending on the level of security the client needs or can afford. However, in security negligence cases for third-party crimes, more detailed crime data, such as incident reports, may be required to establish legal points like the foreseeability of crime.

FORESEEABILITY OF CRIME, A FACTOR in judging whether security is negligent, hinges almost exclusively on the use of official crime statistics to support the circumstances of the case. Historically, victims of crime in public places were obligated to endure their losses unless assailants were caught and forced by the courts to provide restitution. More recently, however, through civil litigation victims are placing more responsibility on various institutions to provide them with reasonable and adequate security. Increasingly, it is held that if a corporate entity--a store, shopping center, hotel, apartment building, park, train station, or airport--invites citizens onto its premises, it must share a portion of the responsibility for the invitees' safety.

Since the 1970s civil litigation alleging negligent security has increased substantially. The 1974 attack on entertainer Connie Francis in her Long Island Howard Johnson motel room is often cited as the case that opened the floodgate to third-party negligent liability. (1) Her negligence suit yielded a $2.7 million verdict. The publicity of the case aroused attorneys and victims to the responsibility of management to provide invitees a reasonable degree of safety.

All such cases hinge on the nebulous concept of foreseeability of crime. In other words, have prior similar crimes occurred at the site or in the immediate area to put owners on notice to provide reasonable and adequate security because such crimes may occur again in the future?

At present, no standard of foreseeability exists to guide security consultants. The conflicting decisions in hundreds of cases throughout the United States add to the confusion. The majority of those cases, however, hinge on official crime statistics. In the final analysis, foreseeability must be considered a statistical concept that indicates the probability of future crimes based on past crime trends and other relevant demographic and environmental data.

Certain types of crimes can probably be classified as unforeseeable. A recent incident at Purdue University in Indiana, in which a college student was shot by her husband on campus, was ruled by the court as not reasonably foreseeable. (2) In In other words, the university could not be expected to protect students from such a random domestic crime. Similarly, assaults and other crimes between tenants have been ruled unforeseeable. [3]

Cases in which mentally ill people commit bizzare crimes in low-crime areas have been ruled by the courts as generally unforeseeable on the part of landlords, fast-food restaurants, and the like. [4] The Stockton, CA, school shooting of several children by a sniper fits this general category.

For those crimes that are deemed foreseeable, chiefly violent, person-oriented crimes, foreseeability often hinges on the number of similar prior crimes at the site in question. Some security consultants have argued that the presence of dissimilar crimes (certain types of property crimes) should forewarn a proprietor that more serious crimes may be forthcoming. However, crime statistics do not generally support that position. For example, UCR data indicate that one aggravated assault can be expected in 1,000 motor vehicle thefts and that one murder can ensue from 45,000 motor vehicle thefts. [5]

While some courts, such as those in Michigan, have been reluctant to hold businesses responsible for protecting their patrons from third-party crimes, [6] most states have wrestled with the question of what number of prior similar crimes is necessary to trigger foreseeability.

Such cases can be broken down into several types: no prior crimes, one to few prior crimes, and several prior crimes. Some recent rulings have held that even a first violent crime on a premises is sufficient for establishing foreseeability and should spur a proprietor to take reasonable precautions to provide protection to the patron. [7] These rulings are probably in the minority and often hinge on the dangerousness of the surrounding area or the special nature of the site in question. Several recent cases involving hospital patients raped by hospital staff fit this category. [8]

A large number of cases involving one or few prior similar crimes can be found in security law litigation. Instances in which one prior crime occurred on the premises also tend to hinge on the special nature of the site in question or the level of crime in the surrounding area. Some courts have operated under the "one free crime" concept, arguing that a business or other establishment is put on notice to protect patrons adequately after one serious criminal incident occurs.

Cases involving several prior crimes present the least difficulty, especially from a plaintiff's position. However, supporting data must be presented not as a simple frequency or total of crimes; it must be understood within the context of trend analysis over time. The analysis and presentation of crime data is vital even in instances where a large number of prior crimes have occurred.

Crime statistics can be used to determine the amount of criminal activity in an area surrounding a crime site. Foreseeability can no longer be analyzed using "prior similars" alone. Other contextual factors in the surrounding environment must be considered: socioeconomic, age, and racial composition; mixture of residential and commercial establishments; land usage patterns; and the like.

No rule of thumb exists for how far to extend the area--that is, one block or a few blocks. Natural geographic boundaries such as rivers or artificial boundaries such as highways may aid in delimitation. In actual practice, some reasonable boundaries must be placed on the parameters of the surrounding area. The use of census tracts allows the consultant to put some closure to a site. The lowest level of census data is the city block or neighborhood, which usually corresponds well to police crime statistics since most police districts are plotted with census maps.

Simply finding a large number of crimes in one neighborhood does not allow the security consultant to argue that it is a high-crime area. Other similar neighborhoods in the city must be analyzed for comparison. With crime statistics and population data for each neighborhood, it is possible to calculate a crime rate per 100 or 1,000 population for each neighborhood.

Even then, extenuating factors must be considered. The presence of large, low-rent housing projects, bars, or public schools can skew crime statistics if they are adjacent to a particular neighborhood. They may affect the criminal geography, or the linkage between crime, physical characteristics, and demographic variables.

DEPENDING ON THE OBJECTIVE OF THE security consultant and the needs of the client, analysis of crime statistics can range from the simple to the sophisticated. Similarly, crime data can be interpreted and presented to strengthen a case depending on whether the client is a defendant or plaintiff.

In working with statistics, both conservative and liberal assumptions can be made about the outcomes of data analysis based on the type of statistical tests employed. A liberal procedure will, for example, find a significant difference between two crime averages (such as armed robberies in two neighborhoods), whereas a conservative procedure will indicate that two crime averages are significantly different only when the averages are far apart. Some knowledge of the intricacies of statistical analysis is vital to security forecasting today.

Manual analysis of crime statistics can help acquaint the user with the data. Pencil-to-paper methods are obviously limited, but they can act as a precursor to more elaborate analysis on personal computers. Most descriptive statistics, such as crime averages, standard deviations, and ranges, can easily be computed with calculators. These techniques allow the practitioner to understand the scope and limitations of the data. However, inferential statistics, which allow the consultant to generalize characteristics from the crime data to a larger population, require computer analysis and specialized knowledge of how to interpret such manipulation.

Statistical tests used to analyze crime data vary in their robustness--the power or the strength of the statistical test and its ability to make inferences or generalizations to regions outside the original crime site. Inferential statistics include the most powerful measures and allow the consultant to predict the likelihood of certain crimes based on past crime trends and such other variables as the degree of target hardening and the number of security officers on-site.

A vast array of statistical software packages are now on the market for the security consultant. For both IBM and Apple users, the Statistical Package for the Social Sciences is an excellent software package for analyzing crime data. It allows for both descriptive and inferential analysis of data. Other spreadsheet programs, though less comprehensive, include Lotus 1-2-3, Quatroy, and Wings, which allow the consultant to categorize and analyze crime data for security cases.

Descriptive statistics transform large groups of numbers into a more manageable form. They allow the consultant to describe, organize, and summarize numerical crime data. A simple frequency distribution allows the consultant to see the most frequent class of crimes at a site. He or she can identify any pattern in the distribution of crimes.

If most of the crimes are concentrated in the middle of the frequency distribution with a few high and a few low scores, then the group of crimes resembles what is called a normal distribution. On the other hand, if the bulk of crimes is concentrated at either the high or the low end of the frequency distribution, then the distribution would be described as skewed. This type of analysis would allow the consultant to see trends or patterns in certain types of crimes as they occur at a specific site or between sites. Frequency distributions can be run by day, week, month, or year to obtain comparisons.

Most groups of crimes possess some degree of variability, that is, the categories of crime differ from one another. A measure of variability simply indicates the degree of dispersion among the set of crimes. If the crimes are highly similar, there is little dispersion. If the crimes are highly dissimilar, there is a high degree of dispersion. In short, a measure of variability does nothing more than indicate how spread out the crimes are.

Descriptive statistics such as the mean (statistical average) and standard deviation (deviation-from-the-mean score) can be used to document variability. When there is extreme variability among crime data, these types of descriptive statistics can be packaged in the form of graphs, tables, and scatter plots for a more easily understood presentation.

Inferential statistics allow the researcher to infer the characteristics of a larger group (such as all convenience stores in a city) from the characteristics of a smaller group (such as a randomly selected sample of convenience stores from the total population of convenience stores in a city). Inferential statistics give the consultant the opportunity, for example, to define the statistical probability that future crimes will occur at other convenience stores with the characteristics of the sample of convenience stores gathered.

Several types of inferential statistical tests allow for the prediction of crime based on the input of specific crime variables into a prediction equation. Such prediction is statistical prediction, and not crystal ball prediction. Even the best statistical analysis of all available data cannot foresee where a particular offender will perpetrate his or her next crime.

A popular type of inferential statistic, multiple regression analysis, allows the consultant to make statistical predictions about the likelihood of certain crime victimizations (referred to as the criterion variable) at a site based on past crime data and other relevant predictor variables. Such predictor variables are well-known to the security specialist and include demographic factors (social class, ethnicity, and age composition of an area), the presence of large public housing units, the presence of certain types of crime-generating businesses (such as bars), commercial and residential density, the degree of public and private security in an area, the degree of target-hardening factors, and other environmental design components, such as lighting and the amount of public versus private space.

The prediction model allows the consultant to estimate future crime vulnerability at a site and to indicate which predictor is the most relevant in reducing such vulnerability. The use of these and other statistics for crime prediction is far from perfect and certainly is not a panacea for security analysts. Such techniques are no better than the quality of data used.

Although social scientists have used advanced inferential statistics for many years to make statistical predictions on various topics, including crime, security consultants have generally not taken advantage of the more sophisticated types of analysis of official crime data. The current demand for security consultants to render expert opinions dictates that new, objective techniques be used in making rational security decisions.

William E. Thornton, PhD, is a criminal justice professor at Loyola University in New Orleans. Ellen McKinnon-Fowler is an urban planner and criminal justice specialist with the New Orleans police department. David R. Kent, CPP, is a security consultant with a background in crime abatement and offense-trend analysis. All three are members of ASIS.

[1] Garzilli v. Howard Johnson Motor Lodges, Inc., 419 F. Supp. 1210 (1976).

[2] Klobuchar v. Purdue University, No. 56A-04-8907-CV-314, Indiana Court of Appeals, Fourth District, April 24, 1990.

[3] Morton v. Kirkland, No. 87-1274, District of Columbia Court of Appeals, May 24, 1989; Tarter v. Schildkraut, No. 36849, New York Supreme Court, Appellate Division, First Department, June 27, 1989.

[4] Davis v. Gomes, No. B034547, Court of appeal of the State of California, Second Appellate District, Division Four, February 22, 1989.

[5] Security Law Newsletter, April 1990, p. 47.

[6] See, for example, Read v. Meijer, No. 104962, Michigan Court of Appeals, May 9, 1989.

[7] Erickson v. Curtis Investment Company, No. C7-88-1177, Minnesota Supreme Court, October 27, 1989.

[8] See, for example, Corker v. Appalachian Regional Healthcare, Inc., No. 5.88-1097, US District Court, S.D. West Virginia, 33 (August 1990), ATLA Law Rep. 264, December 12, 1989.
COPYRIGHT 1991 American Society for Industrial Security
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:use of crime statistics in risk management
Author:Thornton, William E.; McKinnon-Fowler, Ellen; Kent, David R.
Publication:Security Management
Date:Apr 1, 1991
Previous Article:A quantitative tool.
Next Article:Faxpionage!

Related Articles
Diagnosing crime trends.
Following the crime of stalking.
Web-based resources.
Quinlan Publishing starts "Computer Crimes Law Bulletin".
Appendix I--methodology.
Foreword, Crime in the United States, 2004.
Campus crime; legal, social, and policy perspectives, 2d ed.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters