Abstract
This research creates a new criminal profile for burglary by establishing the link among certain offender traits, past criminal behavior, and crime scene features. Utilizing latent class analysis (LCA) to identify underlying groups within the offender and offense characteristics, the relationship between certain offense styles and the most likely offender may then be established. These offense–offender profiles may be used by police to predict traits of an unknown offender based on information from a crime scene alone. Based on a sample of 405 burglaries committed between 2008 and 2009 in Florida, four criminal history groups, four offender types, and four offense styles were identified using LCA. A significant relationship was found among the offense styles and offender trait types, as well as between the offender trait and criminal history categories. This study serves both theoretical and practical purposes, as the findings have important implications for academia and law enforcement alike.
Burglary is one of the most common, yet frequently unsolved crimes across the world (UN Survey of Crime Trends, 2004). In the United States it is estimated that more than 2.5 million burglaries occurred in 2008, though fewer than 12% of the crimes were solved (Federal Bureau of Investigation [FBI], 2008). With an average financial loss of $1,725 per victim and a total social cost estimated at $20,000 per burglary, there are also serious economic consequences to this prevalent offense (Vaughn, DeLisi, Beaver, & Howard, 2008).
Although research has been conducted on ways to decrease burglary rates by deterring or rehabilitating offenders (Hough & Moxon, 1985), few studies have examined new methods to aid police in closing unsolved burglary cases. In these cases, there rarely is an eyewitness, forensic evidence, or probable suspects available to guide police to the burglar (Bennell & Canter, 2002). As a result, cases go cold, victims suffer, and perpetrators are free to reoffend.
It therefore seems that a new tool for solving these crimes is sorely needed by law enforcement, preferably relying only on information that is consistently and readily available at every burglary crime scene encountered. For instance, what if it were possible to predict traits of the most likely offender simply by evaluating the characteristics identifiable at a crime scene? Such a technique is possible and has been used by U.S. law enforcement for more than 35 years. Developed by the FBI in the 1970s, offender profiling (OP) has been used to connect known patterns between various offense styles and traits of the offenders who commit them (see Hazelwood & Burgess, 1987; Ressler, Burgess, & Douglas, 1988). The goal of OP is to establish links between an offender’s behavior at the crime scene and his or her background characteristics by demonstrating that offenders who exhibit a similar crime style (e.g., organized or disorganized) are likely to share similar background traits and behaviors. This tool can then be used by police to narrow down the range of possible suspects by predicting a combination of characteristics that an offender is likely to possess based on elements of a crime scene alone.
Born from a body of knowledge based on agents’ field experience and deductive reasoning, OP has developed into a more systematic method of linking details of the crime committed to the demographic, personality, and/or criminal behavior characteristics of the offender. The apparent success of this technique has resulted in profiles being developed, codified, and adopted by law enforcement for three crimes—serial murder, rape, and arson. These profiles were derived from the combined experiences of two agents in the FBI’s Behavioral Science Unit and interviews with 36 convicted murderers to determine patterns linking the offenders’ traits, motives, psychological states, and the type of crimes they committed (Douglas, Burgess, Burgess, & Ressler, 2006). However, these profiles have come under scrutiny in recent years, as they were created using anecdotal evidence, not quantitative data and statistical methods (Kocsis & Cooksey, 2002). A more statistical approach to OP has recently been adopted by the field (see, e.g., Canter, 1995; Farrington & Lambert, 2000), but statistically based profiling has yet to be applied to burglary in the United States.
Therefore, this research aims to create the first statistically derived profile of burglars within the United States by evaluating the association between certain offender traits and behaviors, and offense features identifiable at the scene of a crime. Through a rigorous quantitative analysis and comprehensive data set, we aim to apply the valuable concepts of OP to the highly undetected crime of burglary.
Burglary Profiling
Although no single definition of OP has been established, the various definitions agree that the goal is to establish links between types of crimes committed and the characteristics of offenders who commit those types of crimes. For example, Hazelwood and Douglas (1980) defined the FBI’s approach to profiling as
an educated attempt to provide investigative agencies with specific information as to the type of individual who committed a certain crime. . . . A profile is based on the characteristic patterns or factors of uniqueness that distinguish certain individuals from the general population. (p. 22)
These individual factors include information about the offender’s personality, age, race, gender, religion, marital status, and other social and behavioral factors (Holmes & Holmes, 1996). These factors are often developed through a deductive process, where crime scenes are studied in extreme detail to develop highly specific and customized profiles applicable only to that case and offender (Holmes & Holmes, 1996).
In a more statistical manner, Canter (1995) summarized OP into an A to C equation, where A represents the actions related to a crime known to the police (e.g., crime location, method of entry, state of scene) and C refers to the characteristics of the responsible offender (e.g., criminal history, identifying traits). This suggests that OP is a scientific method used to draw inferences on offender characteristics using knowledge on the actions related to a crime. This inductive approach to profiling, utilized in the current study, is based on the premise that statistical regularities among certain co-occurring features of crimes exist and may be determined from a database of past offenses. As the features of crimes committed by different individuals are similar, it follows that the offenders who commit the various styles must also share some common underlying traits, which can be generalized to other unknown perpetrators of similar crime styles to aid in their identification (Holmes & Holmes, 1996). Furthermore, as this approach is based on statistical regularities on broader subtypes of offenses and subtypes of offenders, it does not require a profiler to have anything more than general knowledge on human behavior to utilize the research findings (Holmes & Holmes, 1996).
In both cases, the classification of offenders (based on behavioral factors and demographic traits) and the classification of crime scene characteristics are required, with a profile created once the link between the offender and offense subtypes is established. Research on burglary offender and offense subtyping, as well as the linking of offense and offender subtypes, is discussed in the following sections.
Subtyping Burglars
Although no profile has been created for burglary in the United States, related research has shown that different types of burglars and different types of burglaries exist, and therefore profiling this offense may be possible.
For instance, Maguire and Bennett’s (1982) ethnographic study of burglars in England established three distinct groups of burglars based on their criminal histories. The three groups were called low-level amateurs, midl-evel professionals, and high-level professionals, though most burglars were from the low-level or mid-level categories, suggesting little crime specialization.
Vaughn et al. (2008) were the first and only researchers to use a statistical analysis to identify offending groups within a sample of burglars in the United States. Using a latent profile analysis on self-reported criminal history data for 456 adult-aged repeat burglars, four discrete groups were identified: young versatiles, vagrants, drug oriented, and sexual predators. Although this study revealed many insights on burglar subtypes based on latent groups in the offenders’ criminal histories and partial demographics, no data on the burglaries these offenders committed were included in the analysis. Therefore, no linking of offender and offense traits, or “profiling,” was possible.
Linking Offense and Offender Characteristics
Several studies have aimed to establish links between crime scene characteristics and the offenders who commit the crimes (see, e.g., Canter & Fritzon, 1998; DeLisi & Scherer, 2006; Trojan & Salfati, 2011), but few have attempted to link offense features to offender characteristics for burglary. Of the two known studies on the topic, subtypes were generated for either offenders or offenses, but not for both, and neither utilized data from the United States. Still, a relationship between certain offense features and certain traits of the offenders who committed them were discovered in both studies, with many results similar to those found in the present research.
In their study of more than 600 burglaries committed by 244 offenders in Finland, Santtila, Ritvanen, and Mokros (2004) used a principal components analysis to find that certain types of offenses (e.g., basic, spontaneous, suburban) were related to certain burglar traits such as age, employment status, co-offending, visiting the target prior to the crime, and living in the city of the offense. A logistic regression indicated that these offense styles could explain half of the offenders’ variation in age alone, indicating that there are variations in burglar characteristics and the way burglaries are committed, and that meaningful associations exist between the two.
Farrington and Lambert (2007) statistically developed profiles of 400 British burglars and violent offenders using identifying characteristics such as gender, age, race, height, build, hair color, eye color, hair length, accent, living circumstances, and previous criminal activity. These profiles were then tested for their association with certain offense and victim characteristics including crime location, time and season of offense, type and method of entry, and reason for the offense. Significant relationships were found between offender traits and the offense and victim characteristics, illustrating how these links may enable police to identify unknown offenders based on features known at a burglary crime scene. Specifically, offender age was strongly related to type and method of entry, bringing a tool to the crime, and the suspected reason for committing the offense. Furthermore, the results repeatedly suggested that White, male, and older offenders commit offenses in a rather different way from female, non-White, and younger burglars. Perhaps even more intriguing is that hair color, hair length, accent, build, and other physical measures were also found to be associated with specific features of the offenses. Although this research was conducted in the United Kingdom, it paved the way for the present research through the use of statistical analysis to derive and link offender characteristics to offense characteristics.
The present study aims to extend previous research by statistically generating a profile for burglary in Florida. To do this, we utilize a statistical approach called latent class analysis to derive the subtypes of burglar traits and criminal histories and relate them to the offense styles also derived through the technique. Through this, we aim to advance the imperfect but promising policing tool of OP and apply it to burglary in the United States. Not only does this research help eliminate the major critique of OP by utilizing statistical analysis to derive profiles for this costly and prevalent offense, it also expands the field by profiling a new crime type—burglary.
Method
Sample and Data
Data were obtained for 405 solved burglary cases, selected randomly using a police crime analysis program, from about 950 solved burglaries in one county of Florida between 2008 and 2009. 1 The sampled county covers more than 1,200 square miles on the east coast of central Florida and is the home to more than half a million residents in 15 towns and a major city. Within the 405 burglary cases, there are 380 unique offenders and 400 unique burglary locations, indicating that some burglars were repeat offenders and some locations were repeat targets.
Data used in this analysis were acquired from official police records, obtained from several agencies within the state and county. Given that each case in this study was “closed,” as the burglar had been identified and arrested, a great deal of information on the offender was available in each police file. This information includes identifying traits of the burglars, along with offending behavior such as the reported motivation for the offense, co-offending details, and the burglar’s relationship to the victim. The police reports also contain key features of the offense, including the type of dwelling burglarized, method of entry, use of tools, state of the crime scene, occupancy at the time of offense, if the burglary was attempted or successful, the type of items stolen, forensic evidence left at the crime scene, and more.
To obtain the criminal histories of the 380 offenders, the state and local county’s Department of Corrections databases were queried, as these contain information on all offenders who commit a misdemeanor or felony offense in Florida. From this search it was possible to determine each offender’s full criminal history for all crimes committed in any county in Florida. As this county is more than 150 miles from the nearest neighboring state, it is unlikely that the offenders would travel more than 3 hours to commit crimes in other states. For the full list of measures and descriptive statistics, see Table 1. A detailed coding sheet is also available from the authors on request.
Descriptive Statistics for All Offender and Offense Characteristics
Note. Percentages calculated excluding missing values.
Criminal History
Research in OP often includes offenders’ criminal histories in the analyses, as it is believed that patterns exist between offenders’ behavior at a crime scene and behaviors in other aspects of their life, such as prior offenses (see, e.g., Davies, 1997; Horning, Salfati, & Crawford, 2010; Salfati, 2008).
Criminal career
Using the information retrieved from the corrections databases, the existence and length of the burglars’ criminal histories, as well as levels of specialization or versatility in prior offenses, were recorded. The offenders’ previous crimes, or those committed prior to the recent burglary, were assigned to one of the following groups, violent crime, theft, burglary, drug crimes, and other offenses, to correspond with previous burglary profiling research (e.g., Maguire & Bennett, 1982; Vaughn et al., 2008).
It should be noted that some of the burglars’ past crimes were contemporaneous, or committed at nearly the same time as one another. This is particularly relevant for burglary, as most states, including Florida, define the crime as the “unlawful or illicit entry of a dwelling with the intent to commit another offense therein” (Fla. Stat. § 810.02(1)(b)). Although the second offense committed during a burglary is often theft, it need not always be. Burglary can occur if an offender enters a home to commit a sexual offense, destroy property, live illegally, or start a violent altercation. Although this can inflate offenses in the criminal career as compared to the number of “crime events,” these distinct and often quite severe offenses should not be disregarded simply because they occurred at the same date as another offense.
The number of years of offending prior to the recent burglary, as well as the number of offenses committed in each crime category and in total, were also recorded for each offender. Together these measures give an indication of the length and severity of the criminal career prior to committing the burglary in the data set.
Age of criminal onset
Age of onset was calculated by subtracting the offender’s date of birth from the date of the earliest arrest recorded in the official records. This was then trichotomized into early (7.0 to 14.0 years), adolescent (14.1 to 21.0 years), and late onset (21.1 to 65.0 years) classifications, as specified by relevant life-course criminology research (see, e.g., Farrington & Hawkins, 1991; Zara & Farrington, 2009).
Co-offending
Co-offending is the act of committing a crime alongside one or more accomplices (van Mastrigt & Farrington, 2009). Although the causes and consequences of co-offending are not yet fully understood, the mere decision to co-offend has many connotations relating to the age, motivation, and location of the offenders (Bernasco, 2006). Co-offending burglars are generally younger on average (M = 21.9) than solo offenders (M = 26.5), with the bulk of co-offending occurring during adolescence, when peer influence is most influential (van Mastrigt & Farrington, 2009). Solo offenders often burglarize within their own neighborhood and are motivated by material gains (Bernasco, 2006).
In this sample, more than half of the burglars co-offended, a figure that corresponds with other major studies including the Cambridge Study of Delinquent Development, where co-offending was highest among burglars and robbers (Reiss & Farrington, 1991).
Relationship to the victim
This aspect of offending behavior has been understudied for property crimes, yielding few insights regarding the relationship of burglary offenders and their victims. However, as burglary is known to have a serious psychological impact on its victims’ sense of safety and identity, this property crime is increasingly viewed as having an interpersonal dimension (Canter & Alison, 1999; Santtila et al., 2004). Research on other interpersonal crimes, such as murder, indicates there is often a pattern between interpersonal offenders and victims, as females are significantly more likely to be murdered by a current or former intimate partner, whereas males are more likely to be murdered by an acquaintance or friend (Bureau of Justice Statistics, 2006). As more than half of the burglars knew the victim of their offense, there is support for the interpersonal hypothesis, and a potentially important difference between burglars who know the victims they select and those who do not. However, just because the burglar and victim may know each other does not necessarily mean that the case will be any easier for investigators to solve. This information often comes after the offender has been identified and the relationship between offender and victim is able to be established. In the cases where the offender and victim see each other, the victim may recognize the offender but not recall a name or how they met.
Offender Traits
Age
Age is a key aspect in assessing where an offender may be in his or her criminal career and can help determine how offending patterns vary over time (Farrington, 1986). For the burglars in this sample, age at the time of the offense ranged from 11.9 to 63.3 years of age, with the mean at 28.3 years old (SD = 11.4). This coincides with previous burglary research, where burglars were found to be about 27 years old (Bache, Crestani, Canter, & Youngs, 2010; Santtila et al., 2004). This is older than most other offenders, where the average age tends to be 19 to 22 years old, with most desisting between ages 24 to 26 (Farrington, 1986).
Gender
Among the most tested and verified “facts” in criminology is that males offend at higher rates than females, though this discrepancy is even more pronounced for more serious and violent offenses (Blumstein, Cohen, Roth, & Visher, 1986; Steffensmeier, 1993). The gender divide is particularly pronounced for burglary, as past research has found that about 90% of burglars are male, and female burglars tend to start offending later and are more likely to take on assisting roles (Cassel & Bernstein, 2001). Therefore, this important measure was included to determine if gender plays a role in offender subtyping.
Race, height, weight, and hair and eye color
Additional identifying traits of the burglars, including race, height, weight, and hair and eye color, were coded from information provided on the police files to assess whether the characteristic is uniform across all resulting typologies, or if certain traits tend to be associated with certain subtypes. Prior research (see, e.g., Farrington & Lambert, 2000, 2007) has found that relationships exist between these physical traits and certain offense behaviors, though no theory has been proposed as to why such a relationship would occur. It has been suggested that several of these measures may be related by a single overarching factor, such as age, race, or gender. However, as there are not perfect correlations among these variables, as race does not perfectly predict eye color, for instance, then eye color adds to race in helping to identify the type of offender more accurately than race alone.
Additional identifying information is also very important to profiling as the physical traits are often relevant in police investigations and in the identification of offenders. For instance, police often use composite sketches or “be on the lookout” (BOLO) advisories to gain information from the public on potential leads for unsolved crimes. These contain details of an offender’s physical traits, including race, height, weight, and hair and eye color, to identify a potential offender from the general public. It is important to note that this information, along with OP in general, may not be used as evidence or grounds for an arrest, as profiling is simply a tool police may use in prioritizing suspects, generating leads, guiding investigation strategy, and improving the understanding of an offense and likely offender (Jackson & Bekerian, 1997, p. 3).
Offense Characteristics
Motivation
The motivation for committing burglary may be used to differentiate offenders, as burglary is known to be a crime with very different motivations for committing it, often depending on the age of the offender. Most adult burglars engage in the crime for utilitarian purposes, stealing money or goods to support expensive lifestyles or drug and alcohol addictions (Santtila et al., 2004; Wright & Decker, 1994). Younger offenders tend to burglarize for excitement as well as for monetary motivation, whereas others use it as a way to relieve anger or sexual desires, and some just need a vacant home to live in (Bennett & Wright, 1984; Bernasco, 2006; Cromwell, Olson, & Avary, 1991; Vaughn et al., 2008). Based on information recorded by police in the arrest reports such as type and value of items reported stolen, vandalism, and conflict between burglar and victim, as well as the co-occurring offense the burglars were charged with (e.g., grand theft, vandalism, assault), all of the offenders were categorized as having a utilitarian, excitement, or anger/dispute motivation for their burglary.
Although these motivations are inferred from certain key aspects of the burglary, such as type of item stolen and interaction with a victim, until the crime is solved and motivation is determined directly from the offender, it is not possible to establish a motive with absolute certainty. Still, just as a homicide may legally be inferred to have a hate crime motivation based on aspects of the crime scene alone, police will routinely determine motivation for a crime without speaking to a victim or offender directly. Although it is feasible that in some cases these inferences may not be correct, the officers investigating the crimes, and writing the reports used in this study, have a wealth of experience and knowledge regarding the likely motivations for offenses with certain features (e.g., high-value vs. no items stolen, vandalism conducted, etc.). Therefore, the overall likelihood of misclassification of motive is very small, and no changes in suspected motivation versus admitted motivation were found between the initial offense reports and subsequent arrest narratives written by police for this sample.
Time and place of the offense
Although most burglaries now take place during daytime hours when dual-income families are away at school and work (Cassel & Bernstein, 2001), almost as many are also committed during the more stereotypical nighttime hours. Using a daylight calendar that calculates the time of sunrise and sunset for a given date at a specific latitude, the burglary times listed on police reports were dichotomized as occurring either in daytime or at night. The place of the offense was also recorded, using the Uniform Crime Reports coding of residential or commercial dwellings. Although most burglary targets are residential properties, a large proportion are committed against commercial properties (FBI, 2008). Therefore, these offense characteristics were coded from the police reports for analysis.
Scene of the crime
According to one of the original typologies of OP, all crimes and the offenders who commit them are predicted to fall into the organized or disorganized category (Hazelwood & Douglas, 1980). 2 Organized crimes, committed by equally organized and meticulous offenders, are carefully planned, well prepared for (e.g., weapons or tools are brought along), deliberately executed (e.g., potential victims are assessed), and conducted in a methodical manner with any evidence of the offense destroyed or concealed to prevent detection (Kocsis, Irwin, & Hayes, 1998).
Conversely, disorganized offenses are poorly planned or even impetuous, and the offenders often leave a frenzied and chaotic scene of the crime. These offenses are often committed by inexperienced and reckless offenders near their homes, with little effort made to conceal the crime, and a great deal of evidence is typically left behind (Kocsis et al., 1998).
Although this classification has yet to be specifically applied to burglary, measures of planning and level of foresight were coded to take this theory into account. The offense information was taken directly from the police records, as each report detailed whether the burglar used forced or unlawful entry (e.g., finding a door unlocked, window open), if tools were used (e.g., evidence of pry marks from a crowbar), the type and value of items stolen (e.g., petty or grand theft), 3 the state of the crime scene (e.g., if the burglar “ransacked” the scene or left it intact), if forensic evidence or burglary tools were found at the scene, and whether the burglary was successful (e.g., if the burglar escaped after completing the intended task, or if he or she was forced to abort the burglary before completion).
Each of these features are regularly recorded by the police, and therefore the terms reflect the way police code and define them routinely. This increases the practical value of the present research, as there will be little conversion required between the terms utilized in this study and the terms police are familiar with. It is also important to note that each of the features may be identified at the crime scene, without any additional knowledge acquired after the crime is solved.
Analysis
Latent class analysis (LCA) is a “person-focused” method designed to identify latent classes within a group of individuals based on two or more indicator measures (Francis, Bowater, & Soothill, 2004). The goal of this technique is to group subjects into categories where individuals within a group are similar to each other but qualitatively different from individuals in other categories (Muthén & Muthén, 2000).
This method has grown in popularity in criminological research because of its ability to detect underlying patterns, or subgroups, in data (see Besemer, 2012; Deslauriers-Varin & Beauregard, 2010; Vaughn, DeLisi, Beaver, & Howard, 2009; Vaughn et al., 2008; Vaughn et al., 2011). LCA has also been featured in studies of other fields such as neurology, demography, biological psychiatry, and market research (Uebersax, 2009).
LCA has many advantages over related statistical techniques, including cluster, factor, and k-means analyses, as the models rely on person-based case probabilities as opposed to distance or ad hoc probabilities when forming maximum-likelihood-derived classes and the subsequent goodness-of-fit indices (Vaughn et al., 2009). LCA also does not rely on common assumptions that are easily violated in this type of research, such as normality and linearity of the data. However, just as with cluster analysis, the final class solutions of LCA depend entirely on the indicator measures selected to be used in the analysis (Vaughn et al., 2008).
A total of three models were run using Latent Gold v.4.5 software (Vermunt & Magidson, 2005) with the goal of analyzing one to five classes in each. More classes could have been included, but as parsimony is key to any theoretical model, five potential classes were the maximum number selected. Three sets of indicator variables, encompassing all previously mentioned theoretically and practically relevant measures were entered into three separate models for analysis. The goal was to find the optimal class solution for each model, where “all classes are distinct, but where adding an additional class to the model provides no extra explanatory power” (Francis et al., 2004, p. 57).
To determine the optimal final class solutions several goodness-of-fit criteria, including the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Consistent Akaike Information Criterion (CAIC) were examined (Uebersax, 2009). These measures are based on maximum likelihood estimations, providing an indication of the difference between an estimated model and the data observations, much like a chi-square test would. However in LCA, lower values indicate model improvement, as these measures reward model parsimony with lower values. This is particularly true for the BIC and CAIC measures, which tend to favor fewer classes rather than more.
Although there is no definitive statistic for determining the optimal number of classes to select, the model with the lowest values is often chosen, as these measures indicate a better fit to the data than models with higher values (Keribin, 2000). As one model will often not have the minimum values for all three criteria measures, the model with the majority of measures in its favor is typically selected. The goodness-of-fit values, including the log likelihoods (LLs), for all potential class solutions in each of the three models are listed in Table 2.
Fit Indices for All Potential Class Solutions Using Latent Class Analysis
Note. AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; CAIC = Consistent Akaike Information Criterion; df = degrees of freedom; LL = log likelihood; npar = number of parameters. The values in bold represent the best class solution for the data.
We tested one-, two-, three-, four-, and five-group models for each set of analyses, and in all three tests the four-group solution had the best results when parsimony and meaningfulness of the classes were considered. In the criminal history and offense characteristics models, both the BIC and CAIC scores favored the four-class solution, as these scores continued to increase when more groups are added. In the offender traits model, the split between the goodness-of-fit measures indicated that the BIC and CAIC favored parsimony, but the AIC measure suggested two more classes. In this case, the four-group model was selected as it only slightly changed the actual group compositions and served as a compromise between the goodness-of-fit measures. Therefore, it was determined that four classes best fit the data in all three models.
To be sure, bootstrapping was conducted to assess the improved fit between final class solutions and increase validation with respect to the number of classes selected (Vaughn et al., 2009). Bootstrapping is a Monte Carlo resampling technique that relaxes assumptions made about the distribution of variables and performs a large number, 500 in this case, of random iterations to estimate model improvement consequent on the addition of an extra class. In this case, the bootstrapping procedure verified the class solutions identified by the model fit criteria, as four-class solutions showed a statistically significant improvement in fit to the data in all three models, whereas adding a fifth class did not provide a significantly better fit to the data.
After evaluating the model criteria and selecting the optimal number of classes, the contents of the classes were inspected using the conditional item probabilities, which are comparable to factor loadings in factor analysis. These probabilities reflect the likelihood of cases falling into a certain class and help determine the interpretation of each class within the models. The class compositions for the four groups of offender traits, four types of criminal history, and four groups of offense characteristics are discussed in the results section below.
LCA Results
Criminal History Types
The composition of the four classes of criminal history in the final class solution is listed in Table 3, where the percentages of characteristics within each class are also shown. The first latent class within the burglars’ offending histories is characterized by a lack of offending and criminal experience. They make up about 40% of the sample and are referred to as starters. Almost all starters have no prior record, indicating that the burglary was their first known offense. More than half of these burglars are co-offenders, and they tend not to know the victim of the burglary. Age of onset is usually during adolescence, though a notable number had criminal onset later in adulthood (38%).
Comparison of Criminal History and Behavior Across Classes
Class 2 comprises more experienced offenders, as most have a criminal record. Still, they are relatively new to offending, as the majority have been committing crimes for less than 5 years, and most have only committed one or two known offenses. Of the four major crimes evaluated, this group, which composed of 27% of all offenders, is most likely to have previous arrests for burglary and drug crimes. Fewer than 10% committed prior thefts or violent offenses. The age of criminal onset is generally during adolescence or later, with less than 0.2% taking place during childhood. This group slightly favored co-offending and tended to not know the victim. Because of the low levels of offending over the short criminal career, this group is labeled low rate offenders (see Piquero, Sullivan, & Farrington, 2010).
Members of Class 3, the short-term high rate offenders (Piquero et al., 2010), have mostly been offending for the same length of time as the Class 2 low rate offenders, except in this group about 84% have already committed at least three known offenses. Almost all have a criminal record, and nearly all have been arrested for prior burglaries and thefts. About 15% have arrests for drug crimes, but only 0.1% were arrested for prior violent offenses. Age of onset was again likely to occur during adolescence, though almost one quarter began offending during childhood. High rate burglars most often co-offended, though they did not typically know their victim. The high rate offenders made up 14% of the sample.
The final resultant class is distinctive because of the high levels of offending and substantial criminal histories of the constituents. Nearly all the class members have an arrest record, and more than 94% have been offending for more than 5 years. During this time they were arrested for three or more prior crimes, including burglary (36%), theft (38%), drug crimes (39%), and violence (32%). Criminal onset typically occurred prior to adolescence for this group, though a little more than a quarter began offending during adolescence. In the sampled burglary, the majority did not have co-offenders, but they were slightly more likely to know their victim than not. This group, the chronic offenders, composed 19% of the sample.
Offender Trait Types
Results of the offender traits LCA model indicate that four subtypes exist within the data. Class 1 is composed almost entirely of White males who tend to be older in age (65% are older than 25 years old). Members of Class 1 tend to have brown or blonde hair, with brown or blue eyes and an “average to large-sized” build. Only 1% of offenders in this group are short, and nearly 28% are overweight. This group, which makes up 38% of the sample, is referred to as the older White males.
In comparison, Class 2 is made up mostly of people younger than age 25 and has a large composition of female offenders (38%). There is a large proportion of White members, and this class mostly has brown hair. The highest proportion of hazel-eyed offenders is in this group, though most members have brown eyes. This group is notable for being smaller in size, as 65% are short and 84% are thin, and fewer than 2% large or tall. This composition is likely the result of their younger age and fraction of females, and therefore Class 2 is titled the younger Whites. This group is about 26% of the sample.
Class 3 is mostly male, older in age (59% are older than 25 years old), and either Black or Hispanic (78% and 18%, respectively). This class tends to have black or brown hair, with brown eyes and average build. Nearly three quarters are of average weight, and 82% are of average height, and they tend to be on the larger side if they are not average. This group, called the older Black males, makes up about 20% of all offenders.
The final group, the younger minorities, has the highest composition of adolescent offenders (45%), and only one fifth are adults. A relatively high proportion of this group is female (22%), and although nearly 70% of members are Black, this group also has the highest composition of Hispanic members at nearly 30%. This group tends to have dark hair, and nearly all have brown eyes. More than half of the Class 4 group members are short, and 83% are thin. Although nearly 17% are of average or above weight, nearly 50% are of average height and only 0.1% of the burglars are tall. The younger minorities compose 15% of the burglars in the sample. The compositions of these four classes are listed in Table 4.
Comparison of Offender Traits Across Classes
Offense Styles
In the offense characteristics analysis, four latent subtypes were once again identified, listed in Table 5. The first class, the opportunistic style offense, comprised characteristics that indicate that some care was put into the crime, but either foresight or experience was lacking for the crime to unfold in a more professional manner. More specifically, in the opportunistic offense nearly half of the crimes had no forced entry, indicating that the burglars found a door, window, or garage open to enter without force. This is the highest proportion of unlawful entry of all four offense styles. Furthermore, there were uniformly no tools brought by the offenders, and most were motivated by utilitarian interests (73%) or excitement (27%), all consistent with opportunistic offending. These offenses occurred mainly at unoccupied residential dwellings, and slightly more than half occurred at nighttime. These features also suggest this is a crime of opportunity, as the offenders struck while the residents were away, which is a rare occurrence in the evening when residents tend to convene and spend most of their time at home. Still, these offenses show a sense of control as the scene was almost always left in a tidy state, and forensic evidence was left behind just less than one third of the time. However, nearly 40% of these offenses were not successful, indicating that the offenders were interrupted, scared off, or apprehended at or near the crime scene. The opportunistic style makes up 48% of all offenses.
Comparison of Offense Characteristics Across Classes
Class 2 burglaries are highly professional in style, characterized by features indicating that foresight and care were taken to reduce risks and increase gains. About 95% of the offenders brought a burglary tool with them to the crime, indicative of premeditation, and 77% of offenders took their tool away with them afterward. High-value items were stolen in more than 71% of cases, and the motivation for the offense was mostly utilitarian. The scenes were left in a tidy state, and more than half of the offenses showed no signs of forensic evidence. This type of burglary most often occurred at unoccupied residential dwellings during daylight, and the offenders were successful nearly three quarters of the time. This group, the most careful and planned style of burglaries, warranting the title of organized offenses, composed approximately 27% of the burglaries.
Comparatively, Class 3 is characterized by the lack of organization and preparation displayed in the features of the burglary. These crime scenes were almost always left in a state of disarray, with the highest likelihood of evidence and tools left behind (91% and 37%, respectively). Most of these offenses were motivated by excitement; therefore, in most cases no items were actually stolen. Forced entry was most likely to be used on unoccupied targets, with the majority of offenses committed during the day and a substantial number occurring on commercial targets (39%). This distinct style, composing 14% of the burglaries, is labeled the disorganized offense style.
The final class is characterized by the fact that the offense was highly interpersonal, as confrontation often ensued between the burglar and victim. Almost every dwelling was occupied during these burglaries, and nearly 90% of the offenses were motivated by anger or dispute. These burglaries occurred at residential properties at nighttime, and in most cases nothing was stolen. For these types of crimes no burglary tools were brought or used, the scene was tidy, and the offenders escaped successfully. This crime type, called the interpersonal style, was smallest in size, making up about 12% of the burglaries.
Creating a Burglary Profile
After assessing each model and the composition of each of the classes, the relationships among the three sets of subtypes were assessed using contingency analysis in 4 × 4 contingency tables. Chi-square tests of association were conducted to investigate statistical regularities between the categories and determine if the relationships were significantly different from chance. As our goal is to shed light on the most common offender groups within the offense styles and criminal history groups, both the column and row percentages are listed in each of the tables. Column percentages reflect the actual composition of an offense style or criminal history group by the offender types, whereas the row percentages allow for the composition of the offender types within offense style or criminal history to be evaluated against each other.
To determine the extent to which a specific cell within the tables significantly differed from the expected to observed frequency, Adjusted Standardized Residual (ASR) tests were also conducted. 4 ASR values indicate how many standard deviations above or below the expected count an observed count is and signify the importance of the cell to the table’s chi-square value. ASR differs from similar tests of this nature in that it takes into account the overall size of the sample and gives a more fair indication of how much the observed count differs from the expected count (Farrington, Snyder, & Finnegan, 1988). 5 A statistically significant ASR reflects an individual cell that is significantly different from chance expectation. Results of these tests for each relationship are presented in Tables 6 through 8.
Offender Trait Type Versus Offense Style
Note. χ2 = 17.81, p = .037, n = 354, df = 9. Column and row percentages, respectively, are shown below observed cell counts.
Indicates a positive significant Adjusted Standardized Residual (ASR) value at the p = .05 level. bIndicates a negative significant ASR value at the p = .05 level.
Criminal History Type Versus Offense Style
Note. χ2 = 9.42, p = .399, n = 367, df = 9. Column and row percentages, respectively, are shown along with observed cell counts.
Indicates a positive significant Adjusted Standardized Residual (ASR) value at the p = .05 level. bIndicates a negative significant ASR value at the p = .05 level.
Offender Trait Type Versus Criminal History Type
Note. χ2 = 30.93, p = .0001, n = 377, df = 9. Column and row percentages, respectively, are shown along with observed cell counts.
Indicates a positive significant Adjusted Standardized Residual (ASR) value at the p = .05 level. bIndicates a negative significant ASR value at the p = .05 level.
Offense Style and Offender Trait Types
The first chi-square test indicates a moderate and statistically significant relationship between the four offense styles and the four offender trait types (χ2 = 17.81, df = 9, p = .037). Although all four offender types have a large concentration within the opportunistic offense style, the group with the largest proportion within the offense style is the older White males (see Table 6). Nearly 41% of all opportunistic offenses are committed by this group, and almost half of all older White males committed opportunistic offenses. However, as opportunistic offenses are committed in high quantities by all offender trait types, it is important to note that this style is committed in the highest proportion by younger minorities (58%). Among organized offenses, older White males again committed the bulk of this offense style, but the younger Whites group also committed about 23% of this type of crime. A significant, but negative, ASR value was found for older Black males in the organized category, meaning that there were fewer organized offenses committed by older Black males than expected (ASR = −1.96, p < .05). Of the disorganized crimes, only 2 (4%) were committed by younger minority offenders, whereas about 32% were committed by younger Whites. More than 39% of the interpersonal style offenses were committed by older Black males, and almost a third by older White males. The relationship between the older Black male category and the interpersonal offense style produced a positive and statistically significant ASR value, indicating that there were substantially more interpersonal offenses committed by the older Black male group than predicted by chance (ASR = 2.87, p < .05).
Offense Style and Criminal History Type
The general relationship between the offense styles and criminal history types was not shown to be significant by the chi-square test (see Table 7). However, a significant ASR value was found in the high rate/interpersonal cell, as no interpersonal offenses were committed by high rate offenders, which is much less than what was expected to occur (ASR = −2.92, p < .05).
Criminal History Type and Offender Trait Type
The final test examined the relationship between the criminal history subtypes and offender trait groups (see Table 8). Results indicate a strong and highly significant association between the two offender groups (χ2 = 30.93, df = 9, p = .0001). The first criminal history category, the starters, has the highest concentration among older White males, followed closely by younger Whites. Both cells produced significant ASR values, as there are substantially more younger Whites in the starters group than expected (ASR = 2.55, p < .05), but there are significantly fewer older White males observed as starters than predicted (ASR = −2.20, p < .05). Younger minorities also have a significant ASR value, as considerably more are in the starters group than expected by chance (ASR = 2.33, p < .05).
Low rate offenders again have the highest proportion among older White males, as this group has high distributions among all the criminal history types. This aside, the younger Whites have the second largest share within the low rate group at 25%. The high rates are least associated with the younger minorities, and although this criminal history type occurs infrequently among all offender trait groups, it is most associated with the older White males and younger Whites. Chronic offenders are most often older White males, though the largest proportion of the chronic offenders within one trait type is older Black males at 31%. Both of the younger offender groups have the lowest proportion among the chronic offenders, with younger Whites just more than 5%. The older Black males and younger Whites within the chronic group also have statistically significant ASR values, with older Black males observed as chronics substantially more than expected (ASR = 3.04, p < .01), and the younger Whites are classified as chronics far less than predicted (ASR = −3.89, p < .01).
Discussion
This research employed a statistical technique on burglary crime data obtained directly from law enforcement agencies to determine if subtypes exist within offenders’ criminal histories, physical traits, and offense characteristics, and if a relationship exists between certain types of burglars and the various styles of burglaries committed. For if we can reliably connect specific types of burglars to specific types of burglaries, committed at specific places and times, we will gain significant insight on how to identify the perpetrators of this prolific and damaging crime.
Implications of LCA Results
Several categories emerged in each of the analyses, which have implications for applied practice, as well as for life-course theories of criminal behavior such as Moffitt’s (1993) adolescent-limited versus life-course persistent typology. For instance, among the four criminal history groups are the low rate and high rate offenders, indicating that two adolescent onset groups of offenders exist, differing in the rate of offending (D’Unger, Land, McCall, & Nagin, 1998; Nagin & Land, 1993; Piquero, Farrington, & Blumstein, 2007). Therefore, it is possible that both of these groups fit Moffitt’s (1993) adolescence-limited offender type, as both have adolescent criminal onset, limited versatility in offending, and short, though perhaps unfinished, offending careers. 6 Still, it is interesting to note that nearly a quarter of the high rate offenders began offending prior to adolescence, whereas fewer than 1% of low rate offenders began offending in childhood. Combined with their high levels of offending, it is possible that the high rate offenders may actually be chronic offenders at an early stage of their criminal career.
Chronic offenders, who began offending during childhood or adolescence, have many characteristics similar to Moffitt’s life-course persistent offenders as they commit a high number of crimes during their career and show very high levels of versatility in their offending patterns. In contrast, the starters group has no criminal record, and although the majority of the group had an adolescent onset, about a third began offending during adulthood. This mirrors the “adult starters” or “late onset” offender subtype found in previous research (DiLalla & Gottesman, 1989; Kratzer & Hodgins, 1999). Although the idea of late onset offenders has been debated within criminology (see Farrington, Ttofi, & Coid, 2009; Zara & Farrington, 2009), these findings provide some support for the group’s existence. It is of course possible that this group is just more skilled at avoiding detection by law enforcement, but as there are quite a few offenders arrested for the first time as adults, that is not an entirely plausible answer (see also McGee & Farrington, 2010). More research using self-report data should be done to test this, and this topic is discussed in more detail later.
In the offense characteristics analysis, there was evidence of a more organized and disorganized style of offending, providing support for the famous typology used historically for profiling murder and rape crime scenes. The results show that this classification may apply to burglary as well, although the typology must be expanded to account for the two additional subtypes that exist beyond the original organized and disorganized categories. The opportunistic style, which is similar to organized offenses in that care is shown in the crime, differs in that evidence of foresight or planning in criminal behaviors is not present. The interpersonal offense is closer to the disorganized style, as both show higher risks and rates of nonutilitarian motivations. However, unlike disorganized offenses, no disarray was left at these crime scenes, and the targets were often residential and occupied during the offense. In sum, these four groups indicate that the organized−disorganized typology traditionally used within OP research is not actually a distinct typology but rather the two extreme ends of a behavioral continuum with at least two unique classes of behavior in between.
The four offender trait groups are relatively straightforward. Still it is interesting to note that age, gender, and race emerged as the main defining features of the four groups, as indeed was found by Farrington and Lambert (2007). These categories are also simple to interpret and have many practical benefits for crime detection.
Profile Implications
Results indicate that a statistically significant association exists between offender trait groups and offense styles. Opportunistic offenses were committed the most by older White males, though younger minorities showed the highest individual proportion of opportunistic offenses out of all the offender groups. Few disorganized offenses were committed by younger minorities, though nearly a third were committed by younger Whites. Organized offenses were committed most often by the older White males group, whereas the older Black males had the strongest association and positive ASR value for interpersonal style crimes, though they were least likely to commit organized offenses. Farrington and Lambert (2007) found a relationship between burglar traits and offense characteristics similar to those of the current study.
Although the general relationship between offense style and criminal history type was not found to be statistically significant, there were a few clear relationships that emerged. For instance, all four offense styles were committed in a high proportion by the starters, though interpersonal offenses were the most common offense style they committed. Also, no interpersonal offenses were committed by high rate offenders, resulting in a significant and negative ASR value for the cell.
The most significant relationship emerged between the offender trait types and the criminal history groups. Younger Whites and younger minorities were starters more often than expected, whereas older White males were found to be starters significantly less than chance. Low rate offenders were most common among the older White males, with younger Whites close behind. Younger minorities were least likely to be high rate offenders, whereas older White males were the most likely to be. Older Black males had the highest composition among chronic offenders when compared to the other trait group proportions in the category—6 times greater than younger Whites and 3 times higher than younger minorities. However, a note of caution is added for this finding, as co-offending and years of offending are known to have a relationship to age of offenders, and age of onset may be a function of using official records utilized to determine the criminal histories.
Burglary versus Violent Crime Profiles
Although there has been only limited research in developing offense–offender typologies for burglary, there is a body of work examining these relationships for murder, rape, and a variety of other crimes (Douglas et al., 2006; Hazelwood & Burgess, 1987; Hazelwood & Douglas, 1980). These offense–offender typologies, or profiles, seem to indicate that a wide assortment of criminal behaviors can be grouped into disorganized and organized traits. For instance, research on murder typologies generally indicates that the more planned and professional offenses are committed by the older, more experienced, and sophisticated offenders (Hazelwood & Douglas, 1980). This aligns with the current findings for burglars, as the more organized offenses were committed by the older White male category of offenders. Additional similarities in terms of offense behavior were also found between burglary and various violent crimes. For instance, organized murderers often bring their weapon with them to the crime and take it from the scene with them (often because of attachment to the weapon and forethought to avoid leaving behind evidence). The crime scenes are very neat, and the murder is methodically completed, often committed for utilitarian motives or personal gratification. These exact offense traits have been found for rape as well (Hazelwood & Burgess, 1987).
On the other hand, disorganized murders are impulsive, are frenzied, and show little concern to prevent evidence being left behind. Disorganized murderers and rapists either do not bring a weapon or find a weapon of opportunity at the scene, and often leave the weapon at or near the crime scene. These traits appear in disorganized burglaries, as the crimes tend to be frenzied (i.e., scene ransacked or left in disarray), unprepared (i.e., no tool was brought to the scene), and unprofessional (i.e., evidence clearly left behind). Disorganized violent offenders have been shown to be younger and less experienced than their organized counterparts. Similar results were found for burglars, as a large proportion of disorganized offenses were committed by younger White offenders.
Although the current study does not contain data on the personality traits of the burglars, which have been shown to strongly relate to the offense behaviors in violent crimes (Douglas et al., 2006), the similarities between other key aspects of burglary and violent crime typologies suggest that distinctive personality traits may also apply to burglars as well. Furthermore, as profiling is based on the principle that a crime scene is a projection of the offender and his or her underlying personality, there is reason to believe that the same personality traits that result in a certain offense type for one crime will also result in a similar offense type when applied to another crime. Put another way, if personality traits are relatively constant and expressed in all actions a person commits, two different crimes committed by the same offender will tend to be conducted in a similar way. Therefore, as organized violent offenders have been shown to be of higher intelligence, take great care in their personal appearance and belongings, manipulate others for personal gain, maintain relationships and skilled employment, be narcissistic, and generally regard themselves as superior to others, there is reason to believe that organized burglars may be similar to them in many respects. Disorganized violent offenders tend to be of average intelligence, feel alienated from society, hold a poor self-image, and exhibit a more haphazard approach to life, which may apply to the disorganized burglars as well. As this relationship indicates that crimes generally reflect the personality of the criminal, it may also be assumed that burglars who commit interpersonal offenses are aggressive, are combative, and hold a sense of invincibility that promotes highly risky behavior. Opportunistic burglars would then be of modest intelligence and relatively self-aware, but exhibit low self-control, have low self-esteem, and have weak social bonds.
Furthermore, if the personality traits, motivations, and behavior of offenders tend to be held and similarly reflected throughout a variety of dissimilar crime types, there is reason to believe that these traits and behaviors are also held and exhibited by offenders across jurisdictions, and even time. Although social context and technological advances may vary greatly by area (particularly on a global scale) and over time, personality traits, behaviors, and motivations for offending show very little variation either over place or time. For example, burglaries in Alaska may vary from those in Florida, as the amount of daylight, level and type of cover, tools widely available, layout of communities, type of escape routes, and occupancy habits are very different between the two states. This may lead to one type of crime being committed more often in one state (e.g., organized over opportunistic offenses), as the environment is more agreeable to offenders with certain personality traits, motivations, and behaviors than others. Put another way, the number or specifics of certain offense styles may change between place and time, but as the key traits of offenders who commit the crimes are highly similar across place and time, the styles will never be completely different or obsolete.
As these personality traits and other behaviors and motivations of offenders tend to be relatively stable within the offender, though they may change after various experiences and situations (Krueger, 1999; Krueger et al., 1994; McGue, Bacon, & Lykken, 1993), only one crime scene is needed to predict the likely attributes of the criminal. Although profiling has traditionally been applied to serial murder and serial rape cases, this has largely been a product of the jurisdictional restraints and objectives of the FBI’s Behavioral Science Unit, which pioneered the research in the area. 7 Evaluating change or stability of offense behaviors over serial offenses may indicate the degree of consistency exhibited by a specific offender, but if crime is a reflection of the personality traits, motivation, and experiences of an offender at the time of the offense, then a series of offenses may not be necessary to develop the profile of an offender. This fact has recently been noted by profilers, including those at the FBI, with OP applied to a growing list of single offenses (Douglas & Olshaker, 1996).
This study was based on single rather than serial offenses, and a significant relationship was found between certain offender traits and offense styles. However, more research is needed to investigate the consistency of offender traits and offense styles over time.
Limitations
Although the police and criminal history records used in this research are a reliable source of data, there are limitations to any data set resulting from its exclusive use (see Bernasco, 2006). For instance, crimes that have not been detected by police are not included in the criminal history data, and therefore some burglars may be more prolific or have an earlier age of onset than the official data reflect. Whether arrested offenders are comparable to undetected offenders is not entirely clear, although self-report studies (e.g., West & Farrington, 1973) suggest that they are similar in many respects. Specifically, research has shown that the most prolific offenders according to official records are highly similar in many respects to the most prolific offenders according to self-reports (Jolliffe et al., 2003) and that similar conclusions (regarding predictors and correlates of offending) are drawn from both official records and self-report data (West & Farrington, 1973). Farrington and his colleagues (2006) also found that most self-reported offenses of a cohort of males were committed by the convicted males. These studies suggest that an offender caught for one crime has likely committed several other crimes without being detected by law enforcement, and the unsolved crimes are unlikely to be committed by a separate, undetected group of offenders. As there are few practical benefits in generalizing profiles resulting from a sample of solved cases to a population of burglaries that have already been solved, the findings on the similarity between detected and undetected offenders’ behavior and that one offender may be responsible for a string of unsolved offenses provide ample reason to generalize the present findings to undetected burglars as well.
Another limitation lies in the accuracy of data included in the police reports, as some measures were based on inferences from trained and experienced law enforcement agents instead of an objective and measurable trait. One such measure is the offender’s motivation for offending, which was determined largely at the scene of the crime using information available from concurrent offenses committed by the offender during the burglary (e.g., assault, grand theft, vandalism), other crime scene elements, interviews with victims and witnesses, and the officer’s experience with similar cases in the past. For instance, in a burglary case involving grand theft where a high-value item was stolen, no other personal items were touched, and the crime was executed in a very professional and non-thrill-seeking manner, an officer would infer that the offender was most likely driven by a utilitarian motivation. Although the offender may have obtained some excitement in committing the crime and may have decided to steal a high-value item only after entering the dwelling, the final and overarching motivation for offending was utilitarian in nature, as only theft, and no other thrill-seeking behaviors, were engaged in. Unfortunately, there is no research or official data available on the accuracy of the inferred motivation at a crime scene. However, as 100% of the sampled cases showed that the inferred motivation on the offense report was in agreement with the motivation reported by offenders in the arrest narratives, there is reason to believe that the motivation determined by law enforcement at the scene was a highly reliable judgment.
Future Research
The next step in research on profiling offenders using criminal history, offense style, and offender traits would be to replicate the study with a new sample of burglars and use self-reported offending data to determine if those profiles differ from the ones derived in the present research. In addition to the comparative benefits of that research, self-reported offending may shed more details on offender decision making in terms of specialization or versatility, as well as persistence and desistence in offending. Longitudinal research has been the obvious choice to determine length of criminal careers, so future work could aim to undertake that type of research. Measures of psychological traits and details of the burglars’ personal lives, such as marital and employment status, education level, and income, could also shed more light on the nature of the typologies derived in the current study. This information may also provide the details needed to determine more categories within each of the classes and create more specific profiles for this offense.
It would also be advisable to apply LCA to other unprofiled offenses to create profiles for those crimes, gain insight on the latent classes that exist within the offenses, and examine the similarity and differences in behavior exhibited by offenders for otherwise dissimilar crimes. Finally, the implementation of the profiles in the field is ultimately necessary to test the accuracy of the findings in a law enforcement rather than academic setting. This test would indicate how accurate the typologies are in correctly profiling offenses and offenders using a statistical and inductive methodology, whether a new method of creating profiles is necessary, and how successful the proposed profiles actually are in the field, in increasing detection rates for burglary.
Research Implications
Several relationships emerged in the analysis that have the potential to be used in police practice, but there are also implications for developmental and life-course theories—most notably Moffitt’s (1993) adolescence-limited and life-course persistent taxonomy. The strong relationship between offender trait and criminal history subtypes supports Moffitt’s theoretical hypotheses regarding gender and offending (see Moffitt, 1994, pp. 39-40) in that men are more likely to have longer and more prolific criminal careers than women, although it may also be that such a relationship is moderated by relationship status, income level, or personality traits that were not analyzed in this research. Still, the fact that these relationships emerged provides support for aspects of Moffitt’s (1993) theory, including those relating to age and gender, and the subsequent differences in offending.
No significant relationship was found between the criminal history subtypes and offense styles, meaning there is no distinct association between the offending history of a perpetrator and the type of offense committed. This is a rather surprising finding for developmental and life-course criminology, as many researchers assume that aspects of criminal behavior such as years of offending, specialization, and co-offending relate to features of an offense such as preparation level, evidence left behind, items stolen, and motivation for offending. However, the few strong links that were found, such as that starters committed all offense styles whereas high rate offenders were extremely specialized, indicate that the style of offense committed may not be static like a criminal’s signature at an offense, but may actually be evolving with time and experience of offending. Further research is needed to examine this concept in more detail.
Finally, the most important practical implication from this work is that the relationship between offense style and offender trait groups may aid police in identifying the most likely type of suspect from a pool of potential candidates. For instance, if a crime scene reflects a disorganized offense, then the top suspects would be older White males, with younger Whites close behind. A younger minority offender would be a very rare occurrence within this offense style, as would older Black males. Therefore, the profiles can aid in the prioritizing of leads so that the most statistically likely suspects are followed up first by police, which will ultimately save time and costs for law enforcement agencies. And unlike costly and time-consuming forensic analysis, the only cost associated with profiling is a one-time expense for training a member or unit of a police department in these methods and results, which would consequently allow this form of profiling to be done on-site for every burglary case. This tool would be even more valuable if offender traits and offense styles were already stored in a statewide criminal record system that could identify a pool of suspects based on research suggesting who the most likely offender to commit a similar offense style would be.
This is also significant for burglary, as forensic evidence, along with confessions and eyewitness testimony, are currently the most promising sources of identifying unknown perpetrators, yet they are frequently unavailable in burglary offenses (Bennell & Canter, 2002). Consequently, police are often left only with crime scene details to conduct their investigation, posing a considerable challenge and a universally high rate of unsolved burglaries (Grubin, Kelly, & Brunsdon, 2001). Although profiling should only ever be a tool used to derive information about the most and least likely offenders for an offense, prioritize existing leads, and illuminate new investigative strategies that may lead investigators closer to forensic evidence, witnesses, or the responsible offender so the crime may be solved, it is still a highly useful tool for law enforcement to achieve these purposes when no other avenues are available to pursue.
Conclusion
These results are highly relevant for law enforcement and other practical purposes, as they indicate that police may be able to derive information on a likely type of suspect from the offense style witnessed at an unsolved burglary. For instance, a disorganized crime scene would indicate that the most likely offender is an older White male, with the other three subtypes progressively less likely to be the perpetrator. As the method of profiling is inductive, rather than deductive, profile types are created using broader statistical regularities in data of past offenses and offenders. Not every offense will fit the “stereotypical” style, and not every offender will be an identical match to the mold. As these profiles rely on several aspects of the offense and offender to be determined, eliminating or changing only one or two will not prevent law enforcement from correctly assigning the right crime type or identifying the most likely offender. Knowing that one or two offender subtypes have a higher chance of being responsible for a certain style of offense, it is then possible, using the conditional item probabilities, to determine the likelihood that a given offender is an adolescent or an adult, male or female, brown or blonde haired, tall or short, and more. Again, such information would be of no benefit as evidence to arrest a suspect or convict during trial and is still just one tool in law enforcement’s “crime-solving kit,” but it could give police a new perspective on a case when other forms of evidence have gone cold during an investigation.
Through this newly established link between features of an offense and identifying traits and behaviors of an offender, we are able to improve law enforcement’s strategy for targeting unknown offenders and provide an additional tool for police to use in cracking the massive number of burglaries that would otherwise go unsolved.
Footnotes
Authors’ Note:
We would like to sincerely thank the police department that generously supplied the records and data that made this entire project possible. We would also like to thank the anonymous reviewers and editor for their thoughtful and beneficial comments on an earlier draft of this article and the members of the FBI’s Behavioral Science Unit for sharing their knowledge and insights on profiling. Finally, we wish to give special acknowledgment to the Cambridge Trusts and Lucy Cavendish College, University of Cambridge, for the financial assistance to undertake this research.