Learning where to offend: Effects of past on future burglary locations
Introduction
Human behavior in space is remarkably regular and predictable. For example, Song, Qu, Blumm, and Barabasi (2010) showed that knowing a person's recent locations greatly helps to predict their future whereabouts. If such spatial consistency also applies to burglars, locations of their past offenses should inform predictions of future ones. There is pending evidence that this is indeed the case.
More than two decades of research shows that past burglary victimization best predicts future burglary victimization risk (Pease, 1998). This repeat victimization risk peaks immediately after the initial burglary, and usually returns to a baseline level within a few months (Polvi, Looman, Humphries, & Pease, 1991). In most repeat burglary victimizations the same offenders perpetrate both the initial burglary and the follow-up, particularly if the period between both events is short (Everson & Pease, 2001). In other words, repeat burglary victimization typically involves burglars returning to their prior targets.
More recent research shows a phenomenon labeled as ‘near repeat’ (Bowers and Johnson, 2004, Townsley et al., 2003), ‘communicability of risk’ (Johnson & Bowers, 2004a), ‘self-exiting point process’ (Mohler, Short, Brantingham, Schoenberg, & Tita, 2011) and ‘spatio–temporal interaction’ (Grubesic & Mack, 2008). In the wake of a residential burglary, the victimization risk is temporarily increased not only for the initially burgled property, but for nearby homes too. The risk decays both in space and time, typically extending up to a few hundred meters, and persisting for a month or two (Johnson et al., 2007, Ratcliffe and McCullagh, 1998). As with repeat victimization, the same offenders are often involved in both initial and ‘near repeat’ burglaries (Bernasco, 2008, Johnson et al., 2009). Apparently, burglars tend to return to their prior target areas. Furthermore, research on the behavioral consistency of serial offenders (aimed to improve crime linkage techniques in criminal investigations) has shown that the two most consistent and distinctive aspects of the modus operandi of serial burglars are where and when they offend (Markson et al., 2010, Tonkin et al., 2012, Tonkin et al., 2011).
Simultaneously, a formal model of burglary location choice has been developed and has been tested in empirical studies (Bernasco, 2006, Bernasco, 2010a, Bernasco, 2010b, Bernasco and Nieuwbeerta, 2005, Clare et al., 2009, Townsley et al., 2013). The main question that these studies address is how —based on which criteria— offenders decide where to commit a burglary. Applying random utility maximization theory (McFadden, 2001) and discrete choice models (Ben-Akiva & Lerman, 1985), the effects are tested of variables that signal —from an offender perspective—benefits, costs and risks of potential burglary locations.
None of these burglary location choice studies, however, has explored how a burglar's previous burglaries might affect where he strikes next.1 Instead, most studies analyzed the burglaries as isolated events that were unconnected to the past criminal endeavors and experiences of the offenders involved. This might be a serious omission, because the (near) repeat victimization studies cited above suggest that burglars often return to prior targets and target areas to commit subsequent burglaries. Common sense, learning theories, and perspectives from behavioral ecology such as optimal foraging theory (see, Johnson, 2014) suggest that offenders should learn from their criminal experiences. For example, one would expect burglars to return to target areas where they have been successful. One would also expect them to avoid alternatives where they failed in the past —where they did not achieve to enter the property, could not steal any valuable items, or where they were sighted or arrested.
Research on (near) repeat victimization has included theoretical and empirical (e.g., Bernasco, 2008; Johnson, 2014; Johnson et al., 2009) work intended to explain why offenders return to previously victimized locations, but a broader theoretical framework that connects the two lines of enquiry has remained underdeveloped. For example, if the same offenders are responsible for the initial and the subsequent burglary, what made them choose the target in the first place? Under what circumstances will they not return? What other criteria guide their choices? In other words, what is missing from the cited literature is an explicit theoretical model of (repeated) burglary location choice.
A recent study in the Netherlands (Lammers, Menting, Ruiter, & Bernasco, 2015) addressed this issue, using a sample of detected crimes (of any type) and the offenders involved. The areas where offenders had committed prior offenses were included in a model of their subsequent crime location choices. The findings showed that offenders tend to return to previous target areas, in particular to those of recent crimes and especially if the previous crimes were of the same crime type.
The study reported here aims to further explore questions of repeat crime location choices and differs in three key aspects from the study by Lammers et al. (2015). First, it used data from another country (United Kingdom). Second, it used a smaller spatial unit of analysis. Whereas Lammers et al. (2015) used areas with an average population of 7000 residents, the areas used here have an average population of only 1,500, potentially providing for more homogenous spatial units and enhanced ecological validity. Third, and most importantly, whereas Lammers et al. (2015) studied all crime types combined, the present study focused on burglary only. This focus allows for more specificity in the literature review, the theory development, the selection of variables, and in the interpretation of empirical findings.
In sum, we develop explicit hypotheses on how the time and location of past burglaries affect a burglar's subsequent decisions of where to commit burglaries, and integrate the theory that motivates these into a more general model of burglary location choice. Using police-recorded burglary data we rigorously test three hypotheses using discrete spatial choice modeling techniques.
The remainder of the paper is structured as follows. Drawing on the criminological literature on burglary, Section 2 reviews existing theory on how burglars decide where to offend, and develops new theory about how previous burglary locations influence that decision. Section 3 addresses data and method, Section 4 describes the findings, and Section 5 reflects on the results and suggests avenues for future research.
Section snippets
Where to commit burglary
How do offenders decide where to commit a burglary, and how do times, places and other characteristics of their prior burglaries enter into that decision? We start by reviewing the criminological literature, which applies rational choice theory to this topic. We subsequently discuss how crime pattern theory suggests that burglars are most likely to offend within their awareness space, which includes places they have visited in the recent past. Finally, we provide arguments why they are likely
Data and methods
Data were acquired for all detected residential burglaries committed between January 2007 and December 2012 in the West Midlands, a large metropolitan UK police force (see Fig. 1). A residential burglary is the actual or attempted illegal entry into a dwelling with the intent to steal (Bernasco, 2014). A ‘detected’ burglary is one that has been reported to the police, and where at least one person has been charged with committing it. The following variables were available for analysis for each
Findings
The estimated model parameters (odds ratios, confidence intervals, and significance level indicators) are listed in Table A1 in the Appendix. For improved interpretation, this information is presented graphically in Fig. 2. Pseudo R2 values for the discrete choice model are expected to be much lower than those for ordinary least squares models. In this case, according to McFadden's guidance, the observed pseudo R2 of .31 represents an excellent fit to the data (McFadden, 1978: 307).
In Fig. 2,
Discussion
This paper addressed the question of whether having burgled in a particular area, a burglar is more likely to target the same area again, and if so, whether this effect depends on the recency of the prior burglaries and whether it spills over to nearby areas.
Inspired by crime pattern theory it was hypothesized that burglars are more likely to target an area if (1) they have burgled in the area before, (2) the prior burglary was more recently committed, and (3) the area is in the proximity of an
Acknowledgments
This research was supported under Australian Research Council's Discovery Projects funding scheme (project number DP110100100, Understanding How Criminals Decide Where and When to Offend) and under the Netherlands Organization for Scientific Research's Innovational Research Incentives Scheme Vidi (project number 452–12–004). The authors would like to thank West Midlands police, and in particular Chief Superintendent Alex Murray, for providing the data analyzed in this paper, and the Editor and
References (67)
The effect of altitude and slope on the spatial patterning of burglary
Applied Geography
(2012)The space/time behaviour of dwelling burglars: finding near repeat patterns in serial offender data
Applied Geography
(2013)- et al.
Target choice during extreme events: a discrete spatial choice model of the 2011 London riots
Criminology
(2013) - et al.
Discrete choice analysis: theory and application to travel demand
(1985) - et al.
Linking crimes using behavioural clues: current levels of linking accuracy and strategies for moving forward
Journal of Investigative Psychology and Offender Profiling
(2014) Co-offending and the choice of target areas in burglary
Journal of Investigative Psychology and Offender Profiling
(2006)Them again? Same offender involvement in repeat and near repeat burglaries
European Journal of Criminology
(2008)Modeling micro-level crime location choice: application of the discrete choice framework to crime at places
Journal of Quantitative Criminology
(2010)A sentimental journey to crime: effects of residential history on crime location choice
Criminology
(2010)Residential burglary
Where offenders choose to attack: a discrete choice model of robberies in Chicago
Criminology
Go where the money is: modeling street robbers' location choices
Journal of Economic Geography
Statistical analysis of spatial crime data
Handbook of Quantitative Criminology
Effects of residential history on commercial robbers' crime location choices
European Journal of Criminology
How do residential burglars select target areas? A new approach to the analysis of criminal location choice
British Journal of Criminology
Crime location choice
Consistency and specificity in burglars who commit prolific residential burglary: testing the core assumptions underpinning behavioural crime linkage
Legal and Criminological Psychology
Who commits near repeats? A test of the boost explanation
Western Criminology Review
Domestic burglary repeats and space-time clusters: the dimensions of risk
European Journal of Criminology
Prospective hot-spotting: the future of crime mapping?
British Journal of Criminology
Crime pattern theory
Formal evaluation of the impact of barriers and connectors on residential burglars' macro-level offending location choices
Australian and New Zealand Journal of Criminology
Explaining repeat residential burglaries: an analysis of property stolen
Examining the relationship between road structure and burglary risk via quantitative network analysis
Journal of Quantitative Criminology
Population turnover and churn: enhancing understanding of internal migration in Britain through measures of stability
Population Trends
Crime against the same person and place: detection opportunity and offender targeting
Like taking candy. Why does repeat victimization occur?
British Journal of Criminology
Take-the-best in expert-novice decision strategies for residential burglary
Psychonomic Bulletin & Review
Spatial behavior: A geographic perspective
Spatio-temporal interaction of urban crime
Journal of Quantitative Criminology
Ecological analysis of urban offence and offender data
Cited by (81)
Ambient population and surveillance cameras: The guardianship role in street robbers' crime location choice
2021, CitiesCitation Excerpt :High schools, ATMs and banks, and car parks are deemed as public supporting facilities (Baudains et al., 2013; Groff & Lockwood, 2014; Long et al., 2018). Bus stops, transportation hubs, and subway stations are treated as public transport stations (Barnum et al., 2017; Bernasco et al., 2015; Hart & Miethe, 2014; Liu, Zhang, et al., 2017; Summers & Caballero, 2017). The density of road network is used to represent traffic accessibility (Law et al., 2016; Long, Liu, Zhou, et al., 2017; Xiao et al., 2018), and it is measured by the ratio of the total length of road network in a neighborhood to the area of the same neighborhood.
Burglars blocked by barriers? The impact of physical and social barriers on residential burglars' target location choices in China
2021, Computers, Environment and Urban SystemsUnpacking the journey to property crime in Changchun, China
2024, Journal of Investigative Psychology and Offender ProfilingPredicting and analysing initiator crime environments based on machine learning for improving urban safety
2024, International Journal of Architectural Research: Archnet-IJARShould We Stay or Should We Go? A Critical Review of Journey-to-Crime Research
2023, Journal of Police and Criminal Psychology(No) Spillovers in reporting domestic abuse to police
2023, Journal of the Royal Statistical Society. Series A: Statistics in Society