Do burglars choose their targets based on rational considerations? The location of the home? Its surroundings? The security precautions that may or may not be in place? If so, it would make it much easier for police and security companies to design strategies aimed at reducing residents' chances of being burglarized. Residents themselves would also be better equipped to employ the most effective security precautions and modify their actions to minimize the risks.

To find out, my colleagues Gideon Fishman and Simon Hakim, and I carried out a logistic regression analysis, taking into account the location, surroundings, and attributes of a home, as well as the presence or non-presence of a variety of security measures.

Background
Gary Becker, the 1992 Nobel Prize Laureate in Economics, published his seminal work on the rational behavior of the criminals in 1968, a work in which he claimed that a criminal evaluates costs and benefits in choosing his life style and also in deciding whether or not to carry out a particular crime. A decade or so later, my colleague Simon Hakim elaborated on Becker's model by introducing a spatial dimension that states that a burglar's choice of a particular home depends on the expected net return of each potentially available home.

No burglar has ever claimed to have explicitly calculated the net return of a planned burglary, and there is considerable skepticism as to whether burglars actually conduct rational searches. Still, it appears that burglars do implicitly evaluate alternative actions without consciously calculating the net return for each act of burglary. Research results consistently demonstrate that when a large number of burglaries are statistically analyzed, they indeed reveal a maximization of net profits by burglars.

The National Institute of Justice, which is the research arm of the U.S. Department of Justice, commissioned several ethnographic studies in the early 1990s in which burglars were interviewed, followed, and then questioned about the process by which they searched out and chose their targets. (The most notable ethnographic studies are those of Wright and Decker [1994], Rengert and Wasilchick [1985, 1994], Cromwell et al. [1991], and Tunnell [1992]). According to this work, burglars seem to follow a sequential decision process. Burglars choose to operate along familiar routes, which they often use while traveling for work or social purposes. By operating in a familiar area, they minimize the risk of facing unknowns. They also make sure that in case of need, they will have a safe escape route and be able to blend into traffic. Next, the residential burglar selects a middle- or high-income neighborhood, preferably of detached family homes. Burglars appear to be attracted to neighborhoods located within three blocks (a quarter of a mile) of a major arterial route.

Once a neighborhood is chosen, the burglar selects a quiet residential street where the chance of being noticed while entering or exiting the premises is low. A mixed commercial-residential street is less attractive due to the higher volume of pedestrians and motor traffic. Once on the street, the burglar scouts for the specific target. Here, he or she simultaneously considers various factors: the value of the home, its surroundings, whether or not anyone is home. Burglars look very carefully for clues to determine if anyone is in the house or not—28 of 30 burglars stated that they would never purposely enter an occupied residence.

The second consideration of burglars is visibility. The actual illegal entry to most houses takes on average 60 seconds, and not surprisingly, a secluded location increases the probability of a house being burglarized. Interestingly, however, once a burglar locks in on the target, it appears little can be done to stop him or her from actually penetrating the premises. One reason for this, according to the studies, is that many burglars use drugs or alcohol before breaking into a house in order to reduce their anxiety and fear. As a result, situational factors such as the presence of window decals, which are not easily noticeable from the street, have little deterrence value, as does the presence of conventional preventive measures such as dead bolts and window locks.

Our Study
Ethnographic studies in which small a number of active burglars were interviewed and followed up cannot provide generalized conclusions on the entire population of burglars. Intentions expressed by burglars do not necessarily materialize in their real behavior. The implications concerning burglars' behavior are, however, very much strengthened if similar results are derived from both the ethnographic studies and an analysis of cross sectional data gathered from homes that have actually been burglarized.

To determine whether or not a burglar is rational in his or her considerations and operating in a manner that maximizes the net benefit from each burglary we asked the following question: Do a large number of observations provide statistically significant evidence that the following five factors affect the incidence of burglary?

Burglars operate within a three-block corridor of familiar routes where they often travel for work or social purposes. Burglars choose a relatively affluent community. They choose a quiet, residential street. Burglars choose a relatively wealthy home that suggests a lucrative haul. Burglars choose a target that is unoccupied, provides for concealed access, and affords easy escape possibilities.

The Data
The data for our study came from a survey questionnaire sent to the residents of Greenwich, Connecticut. Greenwich is one of the 10 wealthiest communities in the United States. The town is also connected by four major highways to New York City, some 30 miles away, and police records do indeed show that most crimes are committed by transient travelers on the route between New York City and Boston. The questionnaire was mailed to all 22,192 households in Greenwich. Attached was a letter from the chief of police encouraging residents to respond. A news release was also sent by the police chief to the local newspapers stressing the importance of the survey and urging residents to respond. A total of 3,014 households responded, of which 339 had experienced burglary incidents; 88 of the burgled homes had a burglar alarm. Of all respondents, 1,318 had a burglar alarm and 1,696 did not.

Our Model
A logistic regression analysis was employed to estimate the probability of household's being burglarized as a function of the following variables: the location of the home with respect to major arterial routes; and the attributes of the house, including it's immediate surroundings, and the security precautions that were in place.

The dependent variable, Whether or Not a House had been Burglarized, was binomial, with 1 indicating that the home had been burglarized within the previous two and a half years, and 0 indicating that it had not.

The independent variables, House Characteristics, Locational Attributes, and Security Precautions included items such as the following:

• Market Value
• Type of residential unit (detached, townhouse, twin, or apartment)
• Proximity to major thoroughfares and schools
• Location of the home on the street (corner or mid-block)
• Type of street (residential, commercial street, cul-de-sac)
• Number of children in family
• Adjoining properties (wooded area, railroad tracks)
• Existence of a burglar alarm
• Existence of dead bolt locks
• Bars on windows
• Pins in sash windows

The number of children in the family was used as an indicator of occupancy, in that the more children in the family, the more likely it was that somebody was in the house at all times. The location of the house on the street was used as an indicator of visibility. The type of adjoining properties and the existence of such items as dead bolts, bars, and pins were used as indicators of accessibility.

As a category, Security Precautions was further divided into four groups: deterrent measures, managerial measures, preventive measures, and detection measures. Deterrent measures are designed to create the illusion that somebody is in the house even when no one is there, and include having a car parked in the driveway at all times, automatic exterior and interior lighting at night, and a radio and/or TV turned on. A yard sign indicating the presence of a burglar alarm was also classified as a deterrent measure. Managerial measures are also designed to reduce the impression that no one is in the house, and include the suspension of the newspaper and mail and residents asking the police and neighbors to keep an eye on the house while they are away. Preventive measures are aimed at making the actual entry more difficult and time consuming, such as the installation of dead-bolt locks and bars on windows. Detection measures transmit a signal that a burglary attempt or actual entry has occurred. Here, the only such measure is a burglar alarm that transmits a signal to a central station that verifies the alert and then dispatches the alarm to the police.

The Results
According to our analysis, occupancy and visibility were the two statistically significant factors. Houses that had one or more of the following attributes: that they were adjacent to a wooded area, that they were detached family homes, or that they were far from schools had a greater probability of being burglarized. On the other hand, the greater the number of children in the household, the greater the probability that the premises would be occupied, and therefore the lower the chance of burglary. Visibility translates into isolation and seclusion and also relates to a house's proximity to major routes providing burglars with an easy escape. Corner homes and homes close to an arterial roads were significantly more prone to burglary. In addition, the higher the value of the home, the more attractive it was to burglars as well.

Security precautions also had a statistically significant effect on the probability of burglary. Three of the four categories—deterrent measures, managerial measures, and detection measures—reduced the likelihood that a house would or would not be burglarized. Interestingly, preventive measures such as dead bolts did not exhibit significant statistical effect on the probability of burglary. These are items that go largely unnoticed from the street, something that would confirm the view that once a burglar locks in on a target, he or she pursues it vigorously.

Next, we calculated the probability of burglary for both inexpensive and expensive homes. Inexpensive homes were defined as having a market value of $150,000 to $300,000, while expensive homes are those valued at $900,000 and over. Not surprisingly, expensive, detached homes, located close to a major highway, that do not exhibit deterrent, managerial, or detecting precautions, are most vulnerable to burglary. The residence with the lowest probability of being burglarized is an apartment or townhouse, far from a highway entrance/exit, with a relatively low market value.

As for security measures, when a home is protected by an alarm, the probability of burglary is virtually nil. From our results, it was obvious that an alarm system is the single most effective measure to help reduce the probability of burglary. Still, the safest homes are those that combine an alarm system with managerial and deterrent measures as well. This finding holds when controlling for all other variables, such as the property value, the type of unit, and the proximity to highway exit or entrance. Specifically, inexpensive, detached homes, located within one quarter mile from a highway entrance/exit, that employ all three categories of precautions have a 0.032 probability of being burglarized—an 11.72 times lower probability than similar homes that do not maintain any security precautions.

When house characteristics and types of precaution measures are analyzed simultaneously, the houses most prone to burglary are expensive, detached homes, located close to a highway entrance/exit, with no security precautions. The probability for a non-detached home with similar characteristics is substantially lower. Regarding the effect of precautionary measures, in the most prone profile, the combination of managerial precautions and an alarm system reduces the probability of burglary from .567 to .139. If one looks at the same profile for non-detached homes, the probability of burglary decreases from .246 to .039. The same pattern is maintained for homes located in less prone areas, where all the probabilities are lower, but display the same trend. At the same time, the value of the property roughly doubles the risk of victimization when controlling for all other variables.

Conclusions
Looking at the results, one area in which my colleagues and I believe our work could have important implications is the insurance industry. Most homeowner policies offer discount for burglar alarms. Based on our findings, insurance companies could vary this discount according to effectiveness. The closer a home is to major arterial routes, the more effective an alarm system is, the higher the discount. Such discounts would also encourage homeowners to install alarm systems, reducing the company's exposure to loss.

Regardless of the specifics, it is clear that there is some rationale, whether conscious or unconscious, to the manner in which burglars select their targets, information the rest of us can use to defend ourselves against them.

-------------

About the Authors
Dr. Yochanan Shachmurove is a Professor of Economics at the City College of The City University of New York and a visiting professor at the University of Pennsylvania, Economics Department. Dr. Gideon Fishman is the Dean of Graduate School at the University of Haifa. Dr. Simon Hakim is a professor of Economics at the School of Business and Management, Temple University

This article is based on “The Use of Household Survey Data: The Probability of Property Crime Victimization,” which appeared in the Journal of Economic and Social Measurement, Vol. 24, No.1, Lead Article, pp. 1-13, 1998.

More Features