Skip to main content
SearchLoginLogin or Signup

Criminal nomadism: A neglected dimension of spatial mobility in sex offending

Published onJun 03, 2022
Criminal nomadism: A neglected dimension of spatial mobility in sex offending
·

Abstract

Purpose. This study investigates criminal nomadism―an individual’s propensity to engage in continuous or intermittent interurban travel as a way to cope with the consequences of their criminal lifestyle and/or as a strategy to adapt to the reality of being a “career criminal.”

Methods. The criminal-career itinerary across Canada of 448 men convicted of sex offenses was reconstructed through individual interviews and analysis of detailed criminal records. Five distinct components of criminal nomadism (i.e., trips, nodes, paths, range, and mesolevel activity space), inspired by crime pattern theory, are suggested and analyzed.

Results. Results show that criminal nomadism is the reality of young and educated Whites who have a prolific criminal career interspersed with long incarceration sentences. Nomadic offenders did not wander freely and randomly, but rather seemed to be looking for opportunities and privacy. Sex-offending variables did not make a significant contribution to predictions, suggesting that criminal nomadism is more a general offending phenomenon than something specific to sex offending.

Conclusions. This study provides supporting evidence that an extensive criminal career is generally associated with a geographically scattered and nomadic lifestyle. Implications for public policies and future studies are discussed.

KEYWORDS

crime pattern theory, criminal career, criminal nomadism, geographic mobility, sex offending, spatial patterns

1. Introduction

The last 30 years have witnessed a proliferation of legislation dealing with the monitoring and tracking of men convicted of sex crimes and who return to the community. Through the enactment of sex offender registry and notification (SORN) laws, many countries around the world, including the United States and Canada, have made the prevention of sexual crimes a public safety and criminal justice priority (Thomas, 2011). By collecting up-to-date information about individuals convicted of sex offenses, including place of residence, prohibiting them from residing near locations where children congregate, and/or notifying neighbors of their presence, these policies are intended to reduce the risk of sexual reoffending.

Thus far, empirical research, mainly conducted in the United States, has failed to provide any convincing evidence that the implementation of SORN laws has contributed to a reduction of sexual recidivism (see Savage & Windsor, 2018, for a review). Moreover, these laws have been associated with several negative collateral consequences for men convicted of sex offenses; these include increased isolation, limited housing options, ghettoization in socially disorganized neighborhoods, personal victimization, psychosocial repercussions, and problems for family members (see for instance Barnes et al., 2009; Socia, 2011; Tewksbury & Lees, 2006; Tewksbury & Levenson, 2009). All these restrictions and undesired negative impacts not only hinder the reintegration of these post-convicted individuals into society, but also drive them to be residentially mobile, transient, and homeless, or even to go underground (e.g., Levenson, 2018; Tewksbury, 2007). Thus, it has been argued that men convicted of sex crimes represent a very mobile group (Murray et al., 2013; Mustaine et al., 2006).

The high residential mobility exhibited by persons convicted of sex offenses in SORN studies contrasts with their mostly local and geographically bounded criminal mobility reported in traditional journey-to-crime research (see Beauregard et al., 2005, for a review). Seemingly contradictory, these results suggest that one could display relatively low mobility at a microspatial or intraurban level (such as committing crimes mostly near one’s home) while being highly mobile at a mesospatial or macrospatial level (such as committing crimes in different neighborhoods, cities, states, or even countries, due to residential mobility). This hypothesized dual nature of spatial mobility has been overlooked in previous studies, which are mostly dominated by analyses of the distance traveled by individuals from their home to the location of their crimes. This study sought to address this knowledge gap by bringing criminal nomadism into the field of environmental criminology and by relying on this new concept to analyze, over the entire criminal career, the mesospatial or interurban mobility of men convicted for sex offending.

1.1 Interurban spatial patterns of men convicted of sex offenses

At the mesolevel of spatial resolution, research on the criminal spatial patterns of men convicted of sex offenses is scarce. There is some anecdotal evidence that some persons engage in extensive and long-range travel to commit their crimes. Clifford Olson, the infamous Canadian found guilty of numerous sexual murders, once traveled over 5,569 km in two weeks in his search for a suitable victim (Rossmo, 2000). In the United States, Israel Keyes flew from the West Coast to Chicago and then drove around 1,000 miles to Essex, Vermont, to commit two murders and one sexual assault (Peters, 2012). While these narratives are informative and prototypical of the extremely mobile “interurban” individual, they have no scientific value for drawing any conclusions about the spatial patterns of men convicted of sex offenses. Some journey-to-crime studies have reported a small number of persons who traveled over a hundred miles to commit a single sexual crime against unrelated adults (e.g., Davies & Dale, 1996; Warren et al., 1998). Recently, Chopin et al. (2020) noted that 13.9% of the 173 extrafamilial sexual homicides they studied from France had been committed more than 50 km away from the perpetrator's residence. Similarly, Martineau and Beauregard (2016) observed that some Canadians found guilty of sexual murders had been willing to travel considerable distances from home to commit their crime, with the maximum distance being 890 km. It has been suggested that White males (e.g., Warren et al., 1998), motivated by deviant and/or sadistic sexual fantasies (e.g., Dietz et al., 1990), tend to travel longer distances from their residence to commit their crimes. Nevertheless, there is a strong consensus in the journey-to-crime literature that most individuals tend to commit their crimes in relatively close proximity to their home. Because criminal-mobility research has almost exclusively focused on intraurban crime trips over a short period (from minutes to hours), it is not surprising to observe so few interurban spatial patterns among men convicted of sex offenses. In a critical review of this type of research, Michaud (2022) proposed that a better understanding of criminal mobility requires developing alternative measures that consider the multidimensionality of the concept of spatial mobility (i.e., which are not limited to the question of distance traveled) and go beyond the home, the criminal event, and the microlevel of spatial resolution.

On the other hand, research on the mesolevel residential spatial patterns of men convicted of sex offenses is more abundant, although it is limited to analyses of the residential mobility of registered sex offenders (RSOs) in the United States. SORN laws in general, and residence restriction policies in particular, have contributed to substantially reducing housing options for RSOs, amplifying their residential instability, transience, and homelessness. When released from prison, many RSOs are unable to return to their home as a consequence of these restrictions, and thus are forced to move. Studies have shown that this is a fallout experienced by 25% to 42% of RSOs (e.g., Levenson, 2008; Levenson & Cotter, 2005a). Another study showed that in two thirds (64%) of cases, the census tract of the address reported by RSOs at the time of registration differed from the one reported at the time of arrest for a sexual crime (Mustaine et al., 2006). Moreover, it has been estimated that 2-3% of RSOs in the United States are homeless or transient, mostly due to residence-restriction policies (e.g., Harris et al., 2014). Studies have suggested that younger RSOs (Levenson, 2008), minority RSOs and RSOs with a “predator” risk status (Levenson et al., 2015), RSOs not under supervision and RSOs with failure-to-register convictions (Socia et al., 2015), and child-molester RSOs (Rydberg et al., 2014) are more inclined to transiency and residential mobility.

Several concerns and limitations can be raised about existing research on the residential spatial patterns of men convicted of sex offenses. First, most studies have analyzed such patterns as an undesirable corollary of SORN laws rather than as a main subject of interest. The focus has been to show to what degree residential mobility was prevalent and detrimental in RSOs’ post-registration life, not to understand or unravel particular spatial patterns. In addition, many studies arguing that SORN laws have caused more residential mobility and transiency among RSOs lack rigorous randomized experimental design with pre-registration and post-registration measurements. It is still unclear to what extent such residential instability was present among men convicted of sex offenses before the enactment of SORN laws. Also, most of the information used in these studies is drawn from public registries that are afflicted by several problems that can affect the quality of the data, such as double-counting, administrative errors, incomplete or inaccurate residential addresses, data-entry inconsistencies, delays in updating registry information, and missing data on the homeless (Ackerman et al., 2012). RSOs freely acknowledge that information listed about them on registries is often inaccurate or missing (e.g., Tewksbury, 2002). It has been estimated that approximately 16% of all American men convicted of sex offenses and required to register are noncompliant and thus are missing from public registries (Levenson et al., 2010). Finally, these registries emphasize where RSOs sleep, with little or no regard to where they socialize or commit subsequent crimes. Indeed, many questions remain regarding the origin, destination, direction, and range of their intercity trips. Given that so little is known about the mesolevel spatial patterns of men convicted of sex offenses, we are in need of an analytical framework that will deepen our understanding of this phenomenon, such as criminal nomadism.

1.2 Defining criminal nomadism

Spatial mobility can be defined as a change of place from one point to another in a given geographical space. This geographical space can be at the micro- (intraurban), meso- (interurban / intranational) or macro- (international) level of spatial resolution. Some scholars (Canzler et al., 2008; Flamm & Kaufmann, 2006) have suggested dividing the concept of spatial mobility into two dimensions: motility (or mobility potential) and movement (or mobility performance). From their perspective, motility is the abstract and unmappable part of spatial mobility. It characterizes one's capacity (accessibility, networks, resources), ability (knowledge, perceptions, skills), and motivations to move through space (Flamm & Kaufmann, 2006). Owning a car, speaking many languages, being in good aerobic shape, and having a good sense of direction are examples of measurable indicators of motility. On the other hand, movements are the tangible and mappable part of spatial mobility, with an origin, a direction, and one or several destinations. The distances traveled to a workplace, the number of bus stops on the way to school, the area covered during daily-routine activities, and the number of different cities visited in a given trip are examples of quantifiable indicators of movement. Incorporating a temporal dimension to these movements allows for the appreciation of more complex phenomena such as speed and acceleration (Ratcliffe, 2006), rhythm, tempo, and timing (Hawley, 1950), and transiency and nomadism.

Transiency and nomadism are semantically related words describing the propensity to engage in frequent and extensive movements through a given geographical space. In sex-offending research, transiency is regularly conflated with homelessness, and both terms are often employed interchangeably to describe the considerable housing instability of RSOs (Levenson & Vicencio, 2016). Nomadism, for its part, is a word loaded with historical and cultural connotations depicting a much more complex form of movement. Its formal meaning has evolved from a primitive mode of group subsistence to include a distinctive modern individual lifestyle.

Prehistoric hunter-gatherer tribes and pastoralist communities had to travel regularly to find the necessary resources to ensure their survival. This nomadism was a form of ecological adaptation in which movement patterns were influenced by climate, terrain, pasture for livestock herding, and the availability of water and food (Khazanov, 1994). The development of agriculture contributed to the sedentarization of many of these nomadic societies and, much later, the rise of industrialization circumscribed this form of survival nomadism to the most remote parts of the world (Reyes-García & Pyhälä, 2017). In modern days, there is a resurgence of nomadism as a way of life. Globalization, the democratization of travel, and the proliferation of communication technologies have unarguably promoted human spatial mobility. Moving from place to place has never been so easy and has even become an increasingly popular way of life, a phenomenon Duncan et al. (2013) call “lifestyle mobility.” Global nomads (also known as neo-nomads or lifestyle travelers) are the epitome of this emergent perspective (D'Andrea, 2006; Kannisto, 2016). These individuals choose a mobile and boundless life, making their living along the way in the various cities, states, and countries in which they transitionally stay. Nomadism can also be considered an important dimension of other lifestyles, such as the one endorsed by digital nomads working remotely (Thompson, 2019), grey-nomad retirees caravanning across the country (Davies, 2011), urban nomads having opted for a life on the streets (Spradley, 1970), backpackers and hitchhikers (O'Regan, 2013), kinetic elite workers (Costas, 2013), circus performers (Terranova-Webb, 2010), and sailors (Koth, 2013).

In lifestyle mobility, the way of life is chosen, but the inherent mobility is not always planned or desired, given its reliance on opportunities and available resources. One could move following a deliberate volition, but also after a constrained decision, or simply because of an absence of options. Nomads' movements are directed and purposeful; they know where they are going and why (Barfield, 1993). Whether it serves to escape a bitter reality, to look for an alternative existence, to find new or better possibilities, to embark on a quest for independence and self-discovery, and/or to engage in the pursuit of freedom and happiness, mobility still has undeniable adaptative and coping particularities. Nomadism does not preclude a “homing desire” (Blunt & Dowling, 2006), but surely challenges the conception of home as something rooted in one physical, geographical, place. Some people have multiple transitionary anchor points (e.g., an apartment, a friend's house, a parent's residence) in various places, which allow them to feel at home everywhere. Even though nomads tend to move on, rather than move back, revisiting and/or returning to previous “moorings” remains a possibility (Cohen et al., 2015). These location-independent individuals can also develop an ability to be at home in their mobility; in this “home-on-the-move” (Germann Molz, 2008), the act of travel provides a form of stability allowing the creation of a sense of belonging similar to home. Lifestyle mobility is distinguished from permanent migration, which occurs at a specific moment in a lifetime and involves a complete and long-lasting relocation of a main residence. It also differs from more temporary forms of spatial mobility such as those exhibited by seasonal migrants and temporary workers, for whom movements do not represent a transition in the life course and are contingent upon returning to the original home.

Criminal nomadism is a new concept that fits well into the perspective of lifestyle mobility. Choosing the criminal lifestyle comes with a set of attributes, motivations, and behaviors in which ongoing spatial mobility occupies a prominent place. Men who persist in crime are characterized by traits—such as impulsivity and low self-control (Gottfredson & Hirschi, 1990), sensation seeking and risky behaviors (Zuckerman, 2007), and present-centeredness (Cornish & Clarke, 2008)—that make them more prone to short-term and “thrilling” nomadism than to long-term and “boring” sedentariness. In addition, a criminal career is typically punctuated by a succession of incarcerations and releases, and/or of long sentences of incarceration, which might induce alienation and stigmatization (Pager, 2003; Schnittker & John, 2007), weaken conventional social ties (Sampson & Laub, 1995), and create a dynamic of carceral impoverishment (Marchetti, 2002). Besides increasing the risks of condition such as poverty, homelessness, drug addiction, and mental health issues, this impedes one's rooting potential and overall stability.

Individuals convicted of sexual offenses in general, and sexual offenses against children in particular, are one of the most hatred and vilified group of people in society (Seto, 2008). A contagious moral panic has resulted in their symbolizing a contemporary form of “folk devils” that nobody wants in their backyard (Cohen, 2011). The hardships associated with the sex-offender label may be a strong motivation to travel to another city in the hope of rebecoming an anonymous citizen. For the criminally driven men wanting to continue committing crimes, being spatially mobile could be a beneficial restrictive deterrence strategy to avoid detection and recognition, and thus limit further apprehensions and convictions (Lammers & Bernasco, 2013). It can also be a tactic to evade police enforcement (Rossmo, 1987; Schwaner et al., 1998), expand their criminal network or prospect for new opportunities (Morselli & Royer, 2008), and/or escape harassment and victimization from other offenders, gangs, or organized crime (Marvell & Moody, 1998). In many ways, ongoing spatial mobility might not only characterize but even serve individuals who persist in crime. In keeping with that line of thought, we define criminal nomadism as an individual's propensity to engage in continuous or intermittent interurban travels as a way of coping with the consequences of their criminal lifestyle and/or as a strategy to adapt to the challenging reality of being a “career criminal.”

1.3 Theorizing criminal nomadism

the novelty of our formulation of criminal nomadism, crime pattern theory (Brantingham & Brantingham, 1993) could provide a relevant and valuable framework for measuring and analyzing this concept. Crime pattern theory is a meta-theory in environmental criminology that combines features from the routine activity approach (Cohen & Felson, 1979; Eck, 2003; Felson, 2006; Felson & Boba, 2010), the geometry of crime (Brantingham & Brantingham, 1981, 2016), and the rational choice perspective (Cornish & Clarke, 1986, 2014). It postulates that each offender has an activity space (i.e., a physical environment) in which most of their customary trips and activities are carried out. This activity space is circumscribed by edges (or physical and perceptual boundaries) and contains routinely frequented nodes (or centers of activity) interconnected by habitually traveled paths (or routes), thus composing an offender’s awareness of space (i.e., what they know about their physical environment).

Crime pattern theory suggests that crimes are most likely to occur in areas familiar to the offender―where they have better practical knowledge―than in comparable unexplored or foreign environments. Functional awareness of a given area allows offenders to minimize the effort required to locate desirable criminal opportunities and to reduce the risks associated with criminal involvement (e.g., familiarity with routes, better knowledge of opportunities, ability to stay inconspicuous), thus representing a more cost-effective decision. Spatial exploration, where an offender ventures into unknown territories to commit crimes or to search for new opportunities, is believed to be a rare phenomenon (Rengert & Wasilchick, 2000). Crime pattern theory also argues that instances of crime are not randomly geographically distributed but rather tend to cluster in certain places (Brantingham & Brantingham, 1995). Some of these areas are called crime generators because the large number of people (criminally motivated or not) who converge there contributes to a high volume of criminal opportunities (e.g., shopping malls, sports stadiums, transportation hubs). Other areas are considered crime attractors because they draw motivated offenders due to their reputation for providing opportunities for crime (e.g., bar districts, drug-selling spots, prostitution areas). By allowing the awareness of space of offenders to intersect with that of potential victims, these places facilitate the creation of crime hot spots and crime corridors (Brantingham et al., 2020). In contrast, crime detractors are places that discourage offenders and offending; these include areas with few attractions or easy natural surveillance (Kinney et al., 2008).

Crime pattern theory emphasizes the short-term (daily, weekly) intrapersonal stability of activity nodes and paths at an urban level of spatial resolution. It has been demonstrated that everyday human movements exhibit a high degree of uniqueness (De Montjoye et al., 2013) and spatial and temporal regularity (Gonzáles et al., 2008; Song et al., 2010). Day after day, people tend to visit the same places and travel along the same routes to perform the same idiosyncratic routine activities. Recently, the mostly atemporal application of crime pattern theory in empirical research has been criticized and suggestions have been made to make it more time specific (Newton & Felson, 2015; van Sleeuwen et al., 2021). Indeed, usual travel patterns are dynamic and eventually change, even for the most predictable individual (Mannering et al., 1994). Sooner or later, nodes will disappear, others will emerge, and paths will be redrawn. In certain circumstances, such as moving to another city, changing jobs, or starting an affective relationship, a complete change of activity space can take place, thus compelling the creation of a brand new set of nodes and paths. Awareness of space will adjust as time passes, fading away for the abandoned environments and progressively improving for the newfound territories. Generally, the old routine will continue to influence the new one, at least for some time (Bernasco, 2010; Bernasco & Kooistra, 2010).

However, to analyze criminal nomadism, crime pattern theory needs to be adjusted to a lower level of spatial resolution (from the urban/city scale to the country scale) and a longer observation period of the spatial patterns (from daily/weekly routines to lifetime variations). As suggested by Brantingham and Brantingham (1984), cities as a whole can be considered nodes if they act as the origin or destination of intercity trips. Within this perspective, cities represent distinct and temporary activity spaces in an offender’s lifetime itinerary, and paths become the routes taken during these intercity trips. The distance between the two most distant cities defines the diameter of an offender's range of operation throughout his life, broadening the microlevel notion of home range (Canter & Larkin, 1993). Metropolises or highly urbanized municipalities, with their abundance of attractive criminal opportunities (Felson, 1987), can be considered crime generators. Conversely, small towns or more rural areas, with their few attractions and limited possibilities for anonymity, can be conceived of as crime detractors. This extension of crime pattern theory can be interpreted as a temporal and geographical “zoom out” of its traditional conception and application.

1.4 Current study

The current state of research on the interurban spatial patterns of men convicted of sex offenses is deficient. Existing studies on this topic have provided data that are anecdotal, on outliers (i.e., criminal-mobility research), or limited to the population of RSOs in the United States (i.e., residential-mobility research). The aim of this study is to contribute to the field of environmental criminology by analyzing criminal nomadism during the criminal career of men convicted of sex offenses. In doing so, we look for answers to the question of why some individuals exhibit a propensity to travel at a mesolevel (interurban level) of spatial resolution. Our study stands out by its: 1) proposal of an innovative concept (criminal nomadism) capable of overcoming some of the limitations of previous research; 2) analysis of this new construct in a large sample of men convicted of sex offenses; and 3) consideration of spatial patterns across all of Canada and over the entire criminal career. Specifically, the study seeks to answer the following questions: What is the prevalence of criminal nomadism in the criminal career of men convicted of sex offenses? What are the individual factors associated with criminal nomadism? What are the predictors of interurban travel among men convicted of sex offenses?

2. Method

2.1 Participants

The original sample included 587 adult males who were sentenced to imprisonment for two years or more for at least one sexual crime. These men were interviewed between April 1994 and June 2000 while they were incarcerated at a federal maximum-security correctional institution located in the province of Quebec, Canada. Inmates were detained at this institution for a period of approximately six weeks, during which they underwent various evaluations (e.g., criminological, psychological, psychiatric) before being transferred to another correctional institution better suited to their risk level and criminogenic needs.

For this study, a subsample of 448 men convicted of sex offenses was drawn from the original sample. Since we were interested in the criminal nomadism of these individuals during their entire criminal career, the 139 persons (23.7%) sentenced only once were removed. At the time of interview, participants were between the ages of 18 and 78 years (M = 38.3 years; SD = 11.1). Excluding those incarcerated for a life or indeterminate sentence, they were serving an average prison sentence of 4.2 years (SD = 2.9). Although everyone in the sample had been sentenced on at least two occasions—including at least once for crimes of a sexual nature—only half (49.8%) had accumulated two or more sentencing occasions for a sex-related offense. Among the 448 individuals in the sample, 119 (26.6%) had committed their sexual offense(s) exclusively against children (i.e., one or more victims aged 11 or younger); 85 (19.0%) had committed their sexual offense(s) exclusively against adolescents (i.e., one or more victims aged 12 to 17); 151 (33.7%) had committed their sexual offense(s) exclusively against adults (i.e., one or more victims aged 18 or older); 89 (19.9%) had committed their sexual offenses against victims of at least two of these categories (i.e., “mixed” sexual offenses); and 4 (0.9%) had exclusively committed non-contact sexual offense(s) such as voyeurism, exhibitionism, frotteurism, fetish burglary, or sexually motivated arson. It should be noted that 44 participants (9.8%) could also be considered to have committed a sexual murder, according to the criteria developed by Ressler et al. (1988).

2.2 Procedures

The participation in this study was voluntary. Individuals were informed they would not receive any compensation and/or preferential treatment as a result of taking part in this research project. Still, the participation rate was quite high (93.5%). Once the consent form was signed, participants were met individually for a semi-structured interview lasting approximately three hours. A computerized questionnaire was used to collect information on a wide range of socio-demographic, psychological, psychiatric, criminological, sexological and victimological variables. The mean kappa of this questionnaire is .87, which corresponds to a very strong inter-rater agreement. The information gathered during the interview was then compared to official documents (e.g., police reports, court transcripts, specialized assessments, correctional files). When disparities were noted between the information disclosed by the participant and that found in the official documents, official information was opted for.

In 2007, a geographic component was added to the original database. The fingerprint-based criminal record of each of the 448 participants was provided by the Royal Canadian Mounted Police (RCMP). Each file contains current information about the individual's criminal history in Canada, including all sentencing occasions, dates of convictions, criminal charges, locations of hearings, and court decisions. For each sentencing occasion, the municipality in which the case was heard was considered to be the municipality in which the related crimes were committed. According to the 2016 Canadian census geographic boundaries (Statistics Canada, 2017), there are 5,162 municipalities (also called census subdivisions) in Canada. About 606 of these (11.7%) have a courthouse, a service point or an itinerant court location allowing them to hear criminal cases. These “judicial municipalities” are strategically dispersed throughout the country, to be able to serve all population settlements in Canada. If a crime is committed in a municipality that does not offer such court services, the case will generally be heard in the nearest judicial municipality within the same judicial district. It is these 606 judicial municipalities that can be found in the fingerprint-based criminal records provided by the RCMP. The level of spatial aggregation analyzed in this study is therefore less detailed than a census tract or a municipality, but finer than a census division or a judicial district. With dates and locations, it was possible to reconstruct the criminal career itinerary across Canada of each individual at a mesolevel of spatial resolution (i.e., inter-judicial municipalities).

As a group, the 448 participants have been sentenced 3,427 times in 122 distinct judicial municipalities across Canada. These sentencing occasions occurred between 1946 and 2006, and involved a total of 10,108 criminal convictions, among which 1,923 (19.0%) were of a sexual nature. The fact that more than 80% of the convictions were for non-sexual offenses is another indication that men convicted of sex offenses are much more similar to the general offending population than formerly believed (Lussier & Healey, 2009; Lussier et al., 2005). In Canada, the Sex Offender Information Registration Act (SOIRA) (L.C. 2004, ch. 10), requiring all men convicted of sex offenses to register with the National Sex Offender Registry (NSOR), came into force on December 15, 2004. In our sample, only 84 sentences (2.5%) were pronounced after this date, and of these, only 13 (0.4%) involved sexual crimes. Therefore, the majority of interurban spatial patterns investigated by the current study describe the period before the enactment of SORN laws in Canada. This allowed for the appraisal of criminal nomadism without the potential effects of such policies, thus providing a base rate estimate of spatial mobility not influenced by the specific legislative context of Canada.

2.3 Variables

2.3.1 Dependent variable

We conceived of criminal nomadism as a continuous rather than a dichotomous construct. Accordingly, a scale was created to analyze the interurban spatial patterns of men convicted of sex offenses during their criminal career. The five variables that led to the composition of this scale were inspired by the theoretical framework of the crime pattern theory and are illustrated in an example in figure 1.

[INSERT FIGURE 1 ABOUT HERE]

The first component is the number of trips, which is the number of times an individual has been convicted of crimes in a judicial municipality different from that of the previous sentencing occasion. This variable gives an indication of the movements of individuals during their criminal career: the higher the score, the more often a person has traveled to another judicial municipality to commit his crimes.

The second component is the number of nodes, which corresponds to the total number of different judicial municipalities in which an individual has committed crimes during his criminal career. Although this variable shares some similarities with trips, it differs in several ways by: 1) capturing the diversity of the individual's hunting grounds; 2) incorporating a strategic dimension corresponding to the desire to avoid the municipalities already visited (i.e., detection-avoidance strategy); and 3) exposing an ability to explore or prospect for new territories. Having numerous nodes inevitably implies having at least as many trips, but the opposite is not necessarily true. For example, a man may commit his crimes by repeatedly traveling back and forth between two judicial municipalities he knows well. In this pattern, there are many trips but few nodes.

The third component is the length of the paths (or routes), which corresponds to the sum of the distances traveled during the trips made by an individual during his criminal career. The distances were measured in kilometers using Microsoft Streets and Trips software, from the center to the center of each pair (origin-destination) of judicial municipalities, following the street network’s quickest temporal path.

The fourth component is the range (or range of operation), which corresponds to the as-the-crow-flies distance between the two most distant judicial municipalities in which an individual has committed crimes during his criminal career. Again, distances were measured in kilometers using Microsoft Streets and Trips software from the center of both judicial municipalities.1 While both paths and range are measures of distance, the latter also gives an indication of the geographical concentration of nodes within an offender’s mesolevel activity space. The higher the score, the more geographically dispersed across Canada are the judicial municipalities (nodes) in which a person has committed his crimes.

The fifth and final component is the mesolevel activity space, which is the sum of the area (in km²) of ​​each of the judicial municipalities (nodes) in which an individual has committed crimes during his criminal career. Activity space is also an indication of the size of the nodes in which an individual has committed his crimes, and by extension, their pool of available criminal opportunities. A high activity-space score means that a person has operated in one or more densely populated cities, whereas a low score indicates that they have committed their crimes in one or more small-sized towns. This component is also a way to take into account a form of intraurban nomadism that may be available to offenders in larger cities, but not in smaller towns. Indeed, intraurban movements between neighborhoods of large cities may be treated as criminal nomadism in the same way that interurban travel between smaller towns is. Because our methodology was able to take into account only the latter (i.e., interurban travel), the mesolevel activity space component also acts as a measure of the possibilities associated with the former (i.e., intraurban travel).

To have a more realistic measure of the activity space of individuals in each judicial municipality, the population center, rather than the overall area of ​​the territory (which may include forests, mountains, bodies of water, agricultural fields), was used. According to Statistics Canada (2012), a population center (which replaced the term “urban area” in 2011) refers to an area with a population of at least 1,000 and no fewer than 400 people per square kilometer. All areas outside population centers are considered rural areas, and the combination of population centers and rural areas cover the entire territory of Canada. Unlike the land area, which is the same from year to year, population centers are dynamic and can increase or decrease as time elapses. To assess the extent of these changes, we compared the area of the population center of 860 municipalities in 1991 and 1996, using their Statistics Canada Census Profile. The results showed that 63.1% of these municipalities had the exact same area of population center in 1991 as in 1996 (i.e., no change). For the others, the average variation over five years was ±3.5 km² (SD = 6.6). The area of the population centers was considered “fixed” during the period in which the spatial patterns of men convicted of sex offenses were studied (1946 to 2006), since: 1) Statistics Canada only began to codify the area covered by urban areas or population centers in 1991; 2) the yearly changes in these areas appeared relatively trifling; and 3) having extremely precise data was not of paramount importance in the measurement of this variable. Data from the Statistics Canada Census Profile of 1991 was used in this study. 2

Criminal-nomadism scale. Since our five components of criminal nomadism have different metrics and variance, they were transformed into z-scores for the purposes of standardization and comparison. Each z-score was then summed and divided by five to create the criminal-nomadism composite scale. At one end of the scale continuum, there is the typical “sedentary” offender, who committed all his crimes during his criminal career in a single, small, judicial municipality. At the other end of the scale continuum, there is the typical “nomadic” offender, who traveled often and over long distances during his criminal career to commit his crimes in numerous highly urbanized judicial municipalities geographically dispersed across Canada. Our criminal-nomadism scale exhibited strong internal consistency, with a Cronbach's alpha (α) value of .837.

2.3.2 Independent variables

Twelve independent variables were used in our statistical analyses. These variables were selected on the basis of their theoretical relevance to the explanation of human mobility and their methodological quality (i.e., absence of multicollinearity and redundancy, few missing values, good ecological validity, sufficient base rate). The variables can be divided into four distinct blocks: 1) control variable; 2) individual factors; 3) criminal-career factors; and 4) sex-offending factors. Descriptive statistics of the 12 independent variables are presented in table 1.

[INSERT TABLE 1 ABOUT HERE]

Control variable. Our criminal-nomadism scale was created from data associated with each individual sentencing occasion. Consequently, the fewer occasions on which an individual has been sentenced, the less information we had about him and his spatial behaviors, and thus the more likely he was of scoring low on the criminal-nomadism scale. Controlling for the total number of criminal sentences variable was crucial to avoid generating misleading or biased results. Analyses were carried out using the natural log transformation of this variable to correct for its highly positively skewed distribution.

Individual factors. Five variables composed the block of individual factors. We dichotomized race as White (0 = no; 1 = yes) and language spoken as Mother tongue is French (0 = no; 1 = yes). Canada is a bilingual country with two official languages: English and French. Quebec is the only province that is predominantly French speaking, whereas the other provinces are essentially English speaking. Given that the participants in the present study were recruited in Quebec (and were therefore overwhelmingly francophone), we wanted to assess the possible impact of a language barrier on their mesolevel spatial behaviors across Canada. A third variable is the number of years of schooling and corresponds to the number of completed years in school, as declared by the participant. The last two variables of this block are the longest work experience (in months) and the longest intimate relationship (in months), and are intended to shed light on the stability of the individual's conventional social ties and/or of formal social controls. For the last three variables, multiple imputation (fully conditional specification method with 10 iterations) was used to impute 28, 22, and 21 missing data, respectively.

Criminal-career factors. Three variables constitute the block of criminal-career factors. Early onset of antisocial activities, extensiveness of criminal careers, and severity of incarceration are well-known particularities of individuals endorsing a criminal lifestyle. Age of onset — first crime refers to the age at which the participant claimed to have committed his very first crime (official or self-reported), whatever its type and regardless of its legal consequences. Duration of criminal career (in years) is the number of years elapsed between the date of the first sentence and the date of the last sentence received by the individual, according to their fingerprint-based criminal record. Finally, prison time (in years) is the number of years a participant had been incarcerated during his criminal career. This variable was appraised manually for each of the 448 men included in the sample, after analyzing the court decisions associated with each sentencing occasion in their criminal record. This task was done by hand to identify overlapping prison sentences and distinguish concurrent terms from consecutive ones. For each prison sentence, the time served inside walls (i.e., prison time) and the time on parole was estimated on the basis of the following general principles: 1) sentences of less than 180 days were considered to have been served entirely inside walls (no parole); 2) sentences of 180 days or more for exclusively nonviolent crimes or administration of justice offenses (e.g., failure to comply with conditions, unlawfully at large, failure to appear, breach of probation) were considered to have been served in prison for one third of the term and on parole for the rest; and 3) sentences of 180 days or more involving violent crimes, sex crimes, firearms-related crimes or drug-related crimes were considered to have been served in prison for two thirds of the term and on parole for the rest.

Sex-offending factors. Three variables make up the last block of sex-offending factors. The level of specialization in sex offending was computed by dividing the number of criminal sentences involving at least one sexual conviction by the total number of criminal sentences. The closer the resulting ratio is to 1, the more sexual offending is a part of the criminal dynamic of an individual’s offense history. A common characteristic of these “specialists” in sex offending is the presence of deviant and/or sadistic sexual fantasies (Hanson & Bussière, 1998; Quinsey et al., 1995), something that has also been associated with longer crime trips (Davies & Dale, 1996; Dietz et al., 1990). Total number of sex crime victims tallies all official child, adolescent, and adult victims who had been sexually assaulted by the individual in his criminal career. The last independent variable was dummy-coded to correspond to the types of sexual offenses committed by the participants: sexual offenses exclusively against children (0 = no; 1 = yes), sexual offenses exclusively against adolescents (0 = no; 1 = yes), and sexual offenses exclusively against adults (0 = no; 1 = yes). The category mixed sexual offenses (0 = no; 1 = yes) was considered the reference group.

2.4 Analytical strategy

The study followed a three-step approach to analyze the mesospatial or interurban mobility of men convicted of sex offenses during their criminal career. Descriptive statistics are presented first to reveal the range of scores of the participants on the five components of the criminal-nomadism scale. A hierarchical multiple linear regression analysis was subsequently performed to predict the score on the criminal-nomadism scale from several variables measured at the end of the criminal career. Finally, in order to deepen our understanding of the mesolevel spatial patterns of men convicted of sex offenses, we focused on a more fine-grained and short-term conception of interurban travel. Given the nested structure of our data (sentencing occasions nested in individuals), multilevel modeling was undertaken, to avoid inference and aggregation problems (i.e., the ecological fallacy).

3. Results

Descriptive statistics pertaining to the five components of the criminal-nomadism scale are presented in table 2. Results show that around 29.7% of individuals committed all their official crimes in a single judicial municipality during their criminal career. Only a few (10.9%) engaged in criminal activities in five or more judicial municipalities. Among those who traveled at least once between nodes (n = 315), 47.3% had done so only once or twice. On average, the range of operation of their criminal activities was 257.8 km (SD = 532.2), with the bulk of activities (60.4%) occurring within a range of less than 100 km. It is noteworthy that 12.1% of individuals had perpetrated official crimes in more than one Canadian province during their criminal career. The average mesolevel activity space of individuals, as computed by the sum of the areas of population center of all the judicial municipalities in which they had committed official crimes, was 1,038.1 km2 (SD = 755.0).3 Such a high mean mesolevel activity space is mainly attributable to the fact that 55.8% of the participants had committed crimes at least once in Quebec's biggest municipality, Montreal (area of population center = 1,364.5 km2; Statistics Canada, 1992).

[INSERT TABLE 2 ABOUT HERE]

At first glance, there seems to be an inclination toward sedentariness in our sample of men convicted of sex offenses. The standardized scores on the criminal-nomadism scale translate into a positively skewed distribution displaying a higher proportion of low scores (skewness = .897; SE = .117). However, it is important to keep in mind that some of our criminal nomadism indicators, namely nodes and trips, are substantially influenced by the number of observations (i.e., total number of criminal sentences). For example, men who were convicted on only two different occasions (14.3% of the sample) can have a maximum of 2 nodes and 1 trip; those who were convicted on only three different occasions (10.7% of the sample) can have a maximum of 3 nodes and 2 trips, etc. With the variable total number of criminal sentences also exhibiting a positively skewed distribution, it is not surprising to observe the same pattern in the criminal-nomadism scale. Clearly, our scale tends to overestimate sedentariness and underestimate nomadism among less-frequently-convicted individuals. Further multivariate analyses were therefore needed to control for this bias and to better understand criminal nomadism.

A hierarchical multiple linear regression analysis was conducted to predict the score on the criminal-nomadism scale from several predictors while taking into account the total number of criminal sentences. The assumptions of linearity, no multicollinearity, and normally distributed residuals were verified and fulfilled.4 However, the assumption of the constant variance in the errors was not met. To correct for heteroscedasticity, we carried out a weighted least-squares regression using a standard-deviation function to construct weighted observations (Garson, 2013). Four models were tested, and variables were entered incrementally in four different blocks. Results are presented in table 3.

[INSERT TABLE 3 ABOUT HERE]

Model 1 is intended to account for the effects of the total number of criminal sentences on the prediction. When this variable was entered alone, it unsurprisingly predicted the score of the criminal-nomadism scale (F[1,437] = 231.62, p < .001; R² = .346). The addition of the five individual factors in Model 2 significantly improved the prediction (R² change = .030; F[5,432] = 4.23, p = .001), as did the insertion of the three criminal-career factors in Model 3 (R² change = .025; F[3,429] = 6.09, p < .001). However, the inclusion of the last three sex-offending factors in Model 4 did not significantly improve the prediction (R² change = .005; F[5,424] = 0.76, p = .577). Overall, the four models significantly predicted the score on the criminal-nomadism scale. The full model (Model 4) has the best predictive power (F[14,424] = 20.85, p < .001; R² = .408), which corresponded to a large effect size according to Cohen (1988). The beta weights suggest that the two best predictors (excluding the control variable) were prison time in years (Beta = 0.19; p < .001) and mother tongue is French (Beta = -0.13, p < .001). Sex-offending factors such as specialization in sex offending, total number of sex crime victims, and all types of sexual offenses (i.e., exclusively against children, exclusively against adolescents, exclusively against adults, mixed) were not considered significant predictors.

To this point, criminal nomadism had been analyzed using variables measured or aggregated at the end of the criminal career. However, many of these variables unfold over time. For example, an individual can remain sedentary for several years and then become nomadic. Similarly, he can be very active in sex offending at the beginning of his criminal career and then choose to diversify his criminal activities. In order to take into account the dynamic and longitudinal nature of the criminal lifestyle and its inherent spatial mobility, we focused on the prediction of three components of criminal nomadism―namely trips, nodes, and paths―at multiple points in time (i.e., between each sentencing occasion). The last two components (range and mesolevel activity space) were not considered in these subsequent analyses, given that their theoretical contribution becomes less meaningful and rather extraneous when measured over the short term (i.e., between each sentencing occasion) as compared to the long term (i.e., the entire criminal career). Given the nested structure of the data (i.e., 3,427 sentencing occasions nested in 448 individuals), which violated the ordinary least-squares regression assumption of the independence of observations, multilevel modeling was conducted.

As a first step, our database had to be rearranged in long form to have one row per lower-level unit (i.e., sentencing occasions). Because we focused on the prediction of spatial behaviors between a sentencing occasion A (SO-A; origin) and a sentencing occasion B (SO-B; destination), the row of the last sentencing occasion of each individual was removed. Hence, the 3,427 sentencing occasions become 2,979 origin-destination pairs. As a second step, our independent variables had to be adapted to the multilevel analyses. Regarding the predictors associated with the individual (level-2 variables), we selected only 4 of the 12 independent variables used in the previous hierarchical multiple linear regression analysis, namely White (0/1), mother tongue is French (0/1), number of years of schooling, and age of onset — first crime. These variables were chosen based on the certainty that they had been measured before the first sentencing occasion and because of their stability over time. As for the predictors associated with each sentencing occasion (level-1 variables), seven variables were created. Age at SO-A and progression (in %) of criminal career at SO-A were computed for each sentencing occasion. The progression of the criminal career was calculated by dividing our repeated-measure variable (i.e., the rank of the sentencing occasion in an individual's criminal career) by the total number of sentencing occasions. If the sentencing occasion A (origin) involved one or more convictions for sexual crimes, the variable SO-A involved a sexual conviction (0/1) was coded 1. The judicial municipality related to each sentencing occasion was translated into two dummy-coded variables. Judicial municipality of SO-A has a population > 1,000,000 (0/1) is intended to observe the effect of crime generators such as metropolises or highly urbanized cities on the spatial behaviors of individuals. Similarly, judicial municipality of SO-A has a population < 30,000 (0/1) looks at the effect of crime detractors such as small towns or more rural areas. Finally, we calculated the number of days in prison and the number of days of freedom between each sentencing occasion. If the difference between the number of days in prison and the total number of days between sentencing occasion A and B was 0 or negative (i.e., no freedom time), the observation was removed.5 These cases were considered situations in which the participant was incarcerated while receiving his new sentence(s), either because his arrest and eventual conviction made him available for the police to lay pending charges or because he committed crimes while in prison. A biased estimation of the time the individual truly served in prison versus the time on parole could also be a possible explanation of these “no-freedom” situations. Overall, 516 cases (17.3%) were excluded from further analysis, thus giving a final sample of 2,463 observations. We natural log-transformed the variables days in prison between SO-A and SO-B and days of freedom between SO-A and SO-B to correct for the highly positively skewed distributions. These seven level-1 predictors and four level-2 predictors were used to predict the odds of traveling (related to the notion of trips), the odds of exploration (related to the notion of nodes), and the street-network distance in kilometers (related to the notion of paths) between each individual's sentencing occasions. Three distinct generalized linear mixed models (GLMMs) with a repeated-measures design were performed, using IBM SPSS Statistics version 22.0. GLMM was preferred over other analyses given: 1) its flexibility in dealing with an unbalanced number of observations per participant (in our study, the number of sentencing occasions per individual ranged from 2 to 40); and 2) its capacity to fit non-linear models with outcomes resulting from various probability distributions such as binomial and gamma (Heck et al., 2012). The results are presented in table 4.

[INSERT TABLE 4 ABOUT HERE]

The first model predicts the odds of traveling (0/1) between sentencing occasion A (origin) and sentencing occasion B (destination). We considered that a trip occurred (= 1) when the judicial municipality of origin was different than that of the judicial municipality of destination. Among the 2,463 observations, 39.2% involved traveling. The unconditional-mean model (also called random-intercept-only model) revealed that there was a significant nesting among individuals in the travel or not (0/1) variable, which justified the use of multilevel modeling. The intraclass correlation coefficient (ICC) of the empty model suggested that 23.3% of the odds of traveling between sentencing occasions is explained by between-subject differences and, conversely, that 76.7% was explained by within-subject particularities.6 The addition of the seven level-1 and four level-2 fixed predictors, along with a repeated-measure variable (i.e., the rank of the sentencing occasion in an individual's criminal career), significantly improved the model fit. A likelihood ratio chi-square suggested that the difference of 186.92 (df = 13) in the -2 log-likelihood of the two models was significant at p < .001. A look at the fixed predictors revealed an interesting finding about the dual nature of two time-related variables: the probability of traveling decreases as the person ages (log odds = -0.02; p < .05), and increases as the criminal career progresses (log odds = 0.66; p < .05). These seemingly contradictory findings mean that young people tend to travel more frequently than old people and that those in the early stages of their criminal career tend to travel less frequently than those in the later stages. Individuals convicted for crimes in a metropolis (i.e., with a population of more than 1,000,000) were (1 - exp{-.73}) = 51.8% less likely to travel to another judicial municipality. On the other hand, men convicted for crimes in a small town (i.e., with a population of less than 30,000) were (exp{.36} - 1) = 43.3% more likely to travel. Results also indicate that the likelihood of traveling was positively influenced not only by the log number of days of freedom between sentencing occasion A and sentencing occasion B (log odds = 0.14; p < .001), but also by the log number of days in prison (log odds = 0.07; p < .01). It is noteworthy that a conviction for one or more sex crimes did not seem to impact the propensity to travel to another judicial municipality (log odds = -0.04; p = .73). At the individual level, we observed that more educated, White, individuals who are native English speakers and who exhibited an early onset tended to travel more often than did their counterparts. As for random effects, adjacent sentencing occasions (e.g., 1 and 2, 2 and 3, 3 and 4, etc.) exhibited a mean correlation of .29 (Wald Z = 10.05, p < .001), suggesting that the contribution of autoregression is significant and must be considered in the model’s random-effect covariance structure. Finally, results from Model 1 showed that despite representing an improvement from the null model, there was still a large amount of variability to be explained both within sentencing occasions (variance = .88; Wald Z = 26.04, p < .001) and between subjects (µ0j = .70; Wald Z = 4.65, p < .001).

The second model attempts to predict the odds of exploration (0/1) among the situations in which a trip takes place. We considered that an exploration took place (= 1) when a participant was sentenced in the judicial municipality of sentencing occasion B (destination) for the first time in his criminal career. Among the 965 observations featuring a trip, 62.8% involved the exploration of a “new” judicial municipality. Contrary to model 1, the ICC (0.013) of the unconditional mean model did not provide evidence of significant clustering among individuals with regards to the variable explore or not (0/1). Some authors argue that when the ICC is below the conventional threshold of .05 and the design effect (1 + [average group size – 1] * ICC; see Muthén & Satorra, 1995) is < 2, nesting can be ignored and researchers should consider running traditional one-level regression analysis (Kenny et al., 2006; Heck et al., 2012). However, given the unbalanced number of observations per participant and the non-linearity of our outcome variable, we decided to stick with multilevel logistic modeling. The addition of the fixed effects and the repeated-measure variable did not significantly improve the fit of the model, suggesting that it could be underspecified and/or lacking relevant predictors. Still, two results are of particular interest. First, when a participant was convicted of a sexual crime and decided to travel to another judicial municipality after release, he had (exp{.77} - 1) = 116% greater odds of choosing a judicial municipality in which he had never been convicted. Second, as criminal career progressed, individuals became less adventurous and tended to stick to the same judicial municipalities (log odds = -1.93; p < .001). This means that places visited are not infinite and tended to reach a point of saturation. Even among the most criminally active and spatially mobile individuals, there seems to have been a certain predisposition to stay in known territories.

The third model seeks to predict, among the situations in which a trip took place (i.e., 39.2% of time), the distance (in km) between the judicial municipality of sentencing occasion A (origin) and the judicial municipality of sentencing occasion B (destination). The average distance traveled by individuals between sentencing occasions was 172.2 km (SD = 180.9). Even though this dependent variable had a continuous outcome, a GLMM was preferred to a conventional linear mixed model, because of the highly-positively skewed distribution that required the use of a gamma regression. Consequently, coefficients were interpreted as log means rather than conventional β. The unconditional-mean model showed substantial nesting in our data with regards to the distance traveled by individuals between sentencing occasions, as suggested by the ICC of .460.7 The inclusion of the fixed predictors and the repeated-measure variable significantly improved the model fit (χ2 = 100.14, p < .001). Some predictors exhibited the same pattern as in Model 1. For example, the distance traveled between sentencing occasions decreased as the person aged (log means = -0.02, p < .05), while it increased as the criminal career unfolded (log means = 0.39, p < .05). This means that young people tended to travel over longer distances, just like those in the latter stages of their criminal career. As with Model 1, being sentenced in a metropolis was negatively associated with the outcome variable (log means = -0.21, p < .001), while being sentenced in a small rural town showed a positive relationship (log means = 0.16, p < .05). A noteworthy finding in Model 3 was that having French as a mother tongue decreased the log arithmetic mean outcome by 0.75 (p < .001), which implies that native English speakers tended to travel over longer distances across Canada than native French speakers. Again, the contribution of autoregression is significant in the model (rho = .56, Wald Z = 10.86, p < .001) and a substantial amount of variability remained within sentencing occasions (variance = .72; Wald Z = 10.24, p < .001) and between subjects (µ0j = .40; Wald Z = 4.64, p < .001).

4. Discussion

This study provides supporting evidence that, in men convicted of sex offenses, an extensive criminal career is generally associated with a geographically scattered and nomadic lifestyle. Our analyses showed that the younger someone starts committing crimes, the more their criminal career progresses, the more criminal convictions they accumulate, the more time they spend in prison, and the higher their propensity for intercity travels. These findings link criminal nomadism to important criminal career parameters such as age of onset, persistence, frequency, and incapacitation (Piquero et al., 2003). Specialization and sex-offending variables did not contribute significantly to the prediction of criminal nomadism, suggesting that this is more a general offending phenomenon than specific to sex offending.

Some persons are more “motile” or have higher mobility potential than others. Our results showed that criminal nomadism was more prevalent among participants who were White, young, educated, and native English-speakers. Unlike Levenson et al. (2015), who observed that non-White RSOs were more residentially mobile than White RSOs, we found that White participants displayed a higher propensity for criminal nomadism than non-Whites. This could be attributable to cultural differences in the cognitive structure of space (which may affect their awareness of space and what they know about their environment) and/or to an overall tendency for minorities to become sedentary in certain neighborhoods or in gateway cities (Bauder & Sharpe, 2002). The association between spatial mobility and younger people has been reported in some other studies which linked this to this group’s greater impulsivity, lack of financial resources, higher propensity for homelessness, and fewer conventional social ties (Clarke & Eck, 2003; Levenson, 2008). Research conducted on the general population indicates that the level of education is a significant correlate of individual migration. It has been suggested that a higher level of education not only increases the pool of potential jobs, but also makes someone more efficient in their search for legitimate career opportunities, thus facilitating their decision to move (e.g., David, 1974; Schwartz, 1973). Therefore, it seems that intersectionality matters with respect to spatial patterns of offending (see Paik, 2017, for critical perspectives on intersectionality and criminology). Social inequalities due to differences in race, class, and educational level have been found to limit the propensity of offenders for intercity travel, thus contributing to their sedentariness and possibly also to their ghettoization in socially disorganized neighborhoods.

The finding that native English speakers have a greater predisposition for criminal nomadism than native French speakers must be interpreted cautiously with the particular cultural context of Canada in mind. French is the mother tongue of around 80% of residents of the province of Quebec (where the participants in the present study had been recruited) but only 3% of residents of the other provinces in Canada (Statistics Canada, 2007). This result could be the consequence of a methodological bias specific to this study, which considered people from all Canada (mostly native English speaking) who were sentenced in Quebec, while at the same time neglecting people from Quebec (mostly native French speaking) who were convicted elsewhere in Canada. It is also possible that native French-speakers are confronted with a language barrier which causes their criminality to be geographically bounded to the only predominantly French-speaking province in Canada. This would suggest that the propensity for intercity travel may also be cultural. For an offender, not speaking the main language of a new environment may limit the number of desirable criminal opportunities.

Even though participants were not questioned directly on their motivation to adopt a nomadic lifestyle, our results suggest that those doing so were not wandering freely and randomly but rather seemed to be looking for opportunities and privacy. It has been shown that metropolises (i.e., with a population of more than 1,000,000) tend to pull individuals toward the city, and that small towns (i.e., with a population of less than 30,000) tend to push them away. Indeed, participants convicted of crimes in a metropolis were less likely to travel to another judicial municipality to commit their next crimes, while those convicted in small towns were more likely to travel elsewhere. Our analysis also suggests that those who moved from a metropolis tended to stay rather close to this important node while those moving from a small, possibly remote, rural town may be inclined to travel over longer distances. These results point toward a gravitational-attraction effect of large and populous cities on individuals' criminal behavior: the bigger the mass (i.e., size of the city), the stronger the attractive force, and the farther a person moves from this mass, the weaker its attraction. To our knowledge, this is the first time the magnetic effect of cities on the criminal behavior of individuals has been empirically supported.

Our study has also demonstrated that a conviction for one or several sexual crimes does not significantly increase the overall odds of traveling. Nevertheless, among the men who decided to move after a sexual conviction, there was a strong inclination to head toward “new” judicial municipalities―rather than already visited ones―to commit their next crime(s). This result suggests that a proportion of individuals may be looking for enhanced privacy in a virgin environment after being officially labelled as a “sexual offender” (Levenson & Cotter, 2005b). Moreover, the criminal career of most men is not a linear trajectory of unrelenting illegal activities. It is rather a succession of periods of activity and inactivity that coexists with their fluctuating motivation for crime (Laub & Sampson, 2001). While this study suggests that nomadic offenders are looking for opportunities and privacy, it fails to clarify to what extent this is a need of those inclined to pursue their criminal activities and of those hoping to desist from crime. Despite this study’s focus on the nomadic behaviors of those who persist in crime, the results do not preclude the possibility that participants may have tried at some point to desist by moving to another municipality―even though they ultimately failed by committing a new crime.

The routine-activity approach (Cohen & Felson, 1979; Eck, 2003; Felson, 2006; Felson & Boba, 2010) provides explanations for the attractiveness of large urban cities and the deterrence of small towns for both criminally active and desisting offenders. Metropolises abound with opportunities (both criminal and legitimate) that are linked together by an efficient and ramified street network—what Felson (1987) calls “the sociocirculatory system” of the modern metropolis. This street network provides fast and easy access to opportunities by car or through the use of diverse, accessible, and relatively inexpensive modes of transportation (walking, bicycling, bus, train, subway). Indeed, the high permeability of metropolitan areas facilitates movement between places (Groff et al., 2014). Capable guardianship may also differ depending on the size of the city (Reynald, 2009, 2010; Hollis & Hankhouse, 2019). As compared to smaller rural towns, large cities have quantitatively more guardians against crime, but these guardians may be qualitatively less capable or willing to intervene due to a lack of social cohesion or limited visibility in relevant areas (i.e., the built environment of densely populated cities may hamper a potential guardian's view of what is going on outside their homes) (Avery et al., 2021). Metropolises are also places of high population density, due to residents and visitors (Andresen, 2006; Boivin & Felson, 2018; Felson & Boivin, 2015). This large number of unrelated people clustered in a relatively small geographical area contributes to anonymity and the formation of a significant pool of potential victims and motivated offenders. This spatiotemporal convergence of criminal and legitimate opportunities makes metropolises an optimal place both for active offenders to maintain a prolific criminal career and for desisting offenders to find alternatives.

Privacy is a universal human need that allows people to cope with both social interactions and personal activities (Pedersen, 1997). In the criminal world, two specific forms of privacy are of particular importance, namely reserve (i.e., the reluctance to disclose personal features of self to others) and anonymity (i.e., the state of being among others without drawing their attention and/or being monitored by them). While these are self-explanatory prerequisites for active offenders, whose “survival” is ensured by their ability to avoid detection, the role of privacy and its importance in the post-release life of desisting offenders has been overlooked in criminological studies. In sex-offending research, the desire for anonymity has mostly been associated with the offense-supportive cognitions of individuals who commit online sexual offenses (Paquette & Cortoni, 2021). However, anonymity and reserve have been found to perform an important psychological function of recovery, and are hypothetically related to the types of privacy chosen in response to greater social injuries (Pedersen, 1997) such as the consequences of criminal actions. There is some empirical evidence that, in the absence of privacy, the more someone who has committed a sexual offense fears being devalued, stigmatized and/or discriminated against, the more likely he is to avoid prosocial activities that could help his rehabilitation (Mingus & Burchfield, 2012). Unarguably, the quest for opportunities and the preservation of privacy are among the most important challenges a person convicted of sex crimes will have to face throughout their criminal career. From a rational choice perspective (Cornish & Clarke, 1986, 2014), adopting a nomadic lifestyle could be an accessible and efficient strategy to help overcome these struggles.

5. Conclusion

The paradigm of mobility is shifting. The ontology of sedentarism and its implicit quest for a settled life is no longer the norm, as the valued behavior or the intrinsic life ambition of individuals inhabits our contemporary societies. Being on the move is no more uniquely the unfortunate fate of the disadvantaged or the luxurious privilege of the prosperous. Indeed, nomadism is resurging as a distinctive feature of alternative lifestyles endorsed by a growing minority of pluralistic individuals seeking a better existence or simply coping with the challenges of life. The present study has shown that those who persist in offending or who are “career criminals” are among those for whom ongoing spatial mobility characterizes―voluntarily or involuntarily―their onerous criminal lifestyle. Despite the obvious drawbacks of choosing “routes over roots,” criminal nomadism must be considered a rational behavior with undeniable coping and adaptative capabilities for both active and desisting offenders.

This study has several limitations worth mentioning. To begin with, our sample of men convicted of sex offenses tends to overrepresent the most problematic cases (e.g., federal sentence, high recidivism rate, longer criminal career, high proportion of sexual murders). Also, the reliance on individuals' criminal records for the analysis of their spatial behaviors has engendered some biases. First, these documents only recorded crimes having led to a court judgment of guilt, thus overlooking all the crimes that did not end with an official conviction. Second, we presumed that the judicial municipality in which a person was convicted of a crime was necessarily the judicial municipality in which this crime was committed. Even though the Criminal Code of Canada (R.S.C., 1985, c. C-46) and the Code of Penal Procedure (C-25.1) stipulate that the legal proceedings have to take place in the judicial district where the alleged offenses were committed, a few exceptions exist.8 Third, we presupposed that the chronology of the sentencing occasions was the same as the chronology of the crimes. However, late denunciations, long and complex investigations, and interminable legal proceedings may create situations in which an individual sentenced for a crime is later convicted of an offense committed earlier. Fourth, the level of spatial aggregation given by the judicial municipalities (i.e., around 606 judicial municipalities in Canada) was rather coarse and may hide many interurban trips occurring at a finer level of aggregation (e.g., between the 5,162 municipalities in Canada). This tends to underestimate the prevalence of criminal nomadism in our sample. Despite these limitations, the reliance on individuals' criminal record allowed us to have access to consistent, reliable, and exhaustive longitudinal data that would have been hardly obtainable otherwise. Like many other studies investigating geographic mobility, this research also had to deal with the jurisdictionally bounded nature of our spatial data. By studying the mobility within Canada only of men having been convicted of a federal sentence in Quebec, our data took into account people from all Canada having been sentenced in Quebec, but neglected people from Quebec having been convicted elsewhere in Canada (which represents a highly mobile subpopulation). Finally, the fact that we were unable to control for the location of the offender’s residence at each sentencing occasion and did not question participants directly about their motivations for interurban traveling has constricted the range and the strength of our conclusions.

This research has some implications for public policies. Our results showed that there is a nomadism inherent to the criminal lifestyle that is unaffected by laws intending to control the spatial behaviors of men convicted of sex offenses. While SORN laws have undoubtedly contributed to exacerbating the housing instability of RSOs in the United States (Murray et al., 2013; Mustaine et al., 2006; Turley & Hutzel, 2001; Zandbergen & Hart, 2006), they have manifestly not created the “problem.” Our study also suggests that men convicted of sex offenses may be spatially mobile for many other reasons than the strictures and/or the lack of options caused by these laws. Scholars have considered the residential mobility of these individuals an obstacle to their rehabilitation, without considering the possibility it may actually help them by fulfilling their need for an anonymous “new beginning” and/or by being a desirable component of their nomadic lifestyle. This study highlights the importance of having more pragmatic and evidence-based legislation that puts fewer restrictions on where a person should reside and more constraints on where he should travel or socialize. An indirect collateral consequence of current SORN laws in the United States is the ghettoization of men convicted of sex offenses in socially disorganized neighborhoods (Socia, 2011). This not only creates a false sense of security in “RSO-free” neighborhoods, by underestimating the travel capabilities of those who are motivated to commit sex crimes, but also provides such individuals with nearby criminal opportunities (fewer guardians, many other offenders, more vulnerable victims) while depriving them of accessible legitimate possibilities (licit work, prosocial relationships). Finally, our results suggest that denying the need for privacy of men convicted of sex offenses with notification laws may increase their propensity for intercity travel. Current and upcoming legislation should be adapted to better circumscribe the criminal nomadism of men convicted of sex offenses rather than push them to sedentarize into undesirable and criminogenic neighborhoods.

Future studies should investigate post-convicted individuals' quest for opportunities and need for anonymity through the lens of rehabilitation rather than in terms of a prerequisite for reoffending. This could be accomplished through in-depth qualitative interviews with offenders during which questions should be asked about their motivations for traveling, as well as about how their propensity for interurban travel is influenced by age, race, social class, and criminal stigma. Criminal nomadism should also be studied in other samples―from various countries and different offending populations―to unambiguously confirm the remarkable association we found between an extensive criminal career and a nomadic lifestyle. Once that has been accomplished, it may be time to start to think about nomadism as a new criminal-career parameter.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

All authors certify that they have no affiliations with or involvement in any organization or entity with any interest in the subject matter or materials discussed in this manuscript.

References

Ackerman, A. R., Levenson, J. S., & Harris, A. J. (2012). How many sex offenders really live among us? Adjusted counts and population rates in five US states. Journal of Crime and Justice, 35(3), 464–474. https://doi.org/10.1080/0735648X.2012.666407

Andresen, M. A. (2006). Crime measures and the spatial analysis of criminal activity. British Journal of Criminology, 46(2), 258–285. https://doi.org/10.1093/bjc/azi054

Avery, E. E., Hermsen, J. M., & Kuhl, D. C. (2021). Toward a better understanding of perceptions of neighborhood social cohesion in rural and urban places. Social Indicators Research, 157(2), 523–541. https://doi.org/10.1007/s11205-021-02664-0

Barfield, T. J. (1993). The nomadic alternative. Prentice Hall.

Barnes, J. C., Dukes, T., Tewksbury, R., & De Troye, T. M. (2009). Analyzing the impact of a statewide residence restriction law on South Carolina sex offenders. Criminal Justice Policy Review, 20(1), 21–43. https://doi.org/10.1177/0887403408320842

Bauder, H., & Sharpe, B. (2002). Residential segregation of visible minorities in Canada's gateway cities. The Canadian Geographer/Le Géographe canadien, 46(3), 204–222. https://doi.org/10.1111/j.1541-0064.2002.tb00741.x

Beauregard, E., Proulx, J., & Rossmo, D. K. (2005). Spatial patterns of sex offenders: Theoretical, empirical and practical issues. Aggression and Violent Behavior, 10(5), 579–603. https://doi.org/10.1016/j.avb.2004.12.003

Bernasco, W. (2010). A sentimental journey to crime: Effects of residential history on crime location choice. Criminology, 48(2), 389–416. https://doi.org/10.1111/j.1745-9125.2010.00190.x

Bernasco, W., & Kooistra, T. (2010). Effects of residential history on commercial robbers’ crime location choices. European Journal of Criminology, 7(4), 251–265. https://doi.org/10.1177/1477370810363372

Blunt, A., & Dowling, R. (2006). Home. Routledge. https://doi.org/10.4324/9780203401354

Boivin, R., & Felson, M. (2018). Crimes by visitors versus crimes by residents: The influence of visitor inflows. Journal of Quantitative Criminology, 34(2), 465–480. https://doi.org/10.1007/s10940-017-9341-1

Brantingham, P. J., & Brantingham, P. L. (1984). Patterns in crime. MacMillan.

Brantingham, P. J., & Brantingham, P. L. (2016). The geometry of crime and crime pattern theory. In R. Wortley & M. Townsley (Eds.), Environmental criminology and crime analysis (2nd ed., pp. 117–135). Routledge.

Brantingham, P. J., Brantingham, P. L., Song, J., & Spicer, V. (2020). Crime hot spots, crime corridors and the journey to crime: An expanded theoretical model of the generation of crime concentrations. In K. M. Lersch & J. Chakraborty (Eds.), Geographies of Behavioural Health, Crime, and Disorder (pp. 61–86). Springer. https://doi.org/10.1007/978-3-030-33467-3_4

Brantingham, P. L. & Brantingham, P. J. (1981). Notes on the geometry of crime. In P. J. Brantingham & P. L. Brantingham (Eds.), Environmental criminology (pp. 27–54). Sage Publications.

Brantingham, P. L., & Brantingham, P. J. (1993). Nodes, paths and edges: Considerations on the complexity of crime and the physical environment. Journal of Environmental Psychology, 13(1), 3–28. https://doi.org/10.1016/S0272-4944(05)80212-9

Brantingham, P. L., & Brantingham, P. J. (1995). Crime generators and crime attractors. European Journal on Criminal Policy and Research, 3(3), 5–26. https://doi.org/10.1007/BF02242925

Canter, D., & Larkin, P. (1993). The environmental range of serial rapists. Journal of Environmental Psychology, 13(1), 63–69. https://doi.org/10.1016/S0272-4944(05)80215-4

Canzler, W., Kaufmann, V., & Kesselring, S. (Eds.) (2008). Tracing mobilities: Towards a cosmopolitan perspective. Ashgate. https://doi.org/10.4324/9781315550459

Chopin, J., Caneppele, S., & Beauregard, E. (2020). An analysis of mobility patterns in sexual homicide. Homicide Studies, 24(2), 178–202. https://doi.org/10.1177/1088767919884601

Clarke, R. V., & Eck, J. (2003). Become a problem-solving crime analyst: In 55 small steps. Jill Dando Institute of Crime Science. https://doi.org/10.4324/9781315060965

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates. https://doi.org/10.4324/9780203771587

Cohen, L. E., & Felson, M. (1979). Social change and crime rate trends: A routine activity approach. American Sociological Review, 44(4), 588–608. https://doi.org/10.2307/2094589

Cohen, S. (2011). Folk devils and moral panics. Routledge. https://doi.org/10.4324/9780203828250

Cohen, S. A., Duncan, T., & Thulemark, M. (2015). Lifestyle mobilities: The crossroads of travel, leisure and migration. Mobilities, 10(1), 155–172. https://doi.org/10.1080/17450101.2013.826481

Cornish, D. B., & Clarke, R. V. (1986). Introduction. In D. B. Cornish & R. V. Clarke (Eds.), The reasoning criminal: Rational choice perspectives on offending (pp. 1–16). Springer-Verlag. https://doi.org/10.4324/9781315134482-1

Cornish, D. B., & Clarke, R. V. (2008). The rational choice perspective. In R. Wortley & L. Mazerolle (Eds.), Environmental criminology and crime analysis (pp. 21–47). Willan Publishers.

Cornish, D. B., & Clarke, R. V. (Eds.). (2014). The reasoning criminal: Rational choice perspectives on offending. Routledge. https://doi.org/10.4324/9781315134482

Costas, J. (2013). Problematizing mobility: A metaphor of stickiness, non-places and the kinetic elite. Organization Studies, 34(10), 1467–1485. https://doi.org/10.1177/0170840613495324

D’Andrea, A. (2006). Neo-nomadism: A theory of post-identitarian mobility in the global age. Mobilities, 1(1), 95–119. https://doi.org/10.1080/17450100500489148

David, P. A. (1974). Fortune, risk and the microeconomics of migration. In P. A. David & M. W. Reder (Eds.), Nations and Households in Economic Growth (pp. 21–88). Academic Press. https://doi.org/10.1016/C2013-0-10564-8

Davies, A. (2011). On constructing ageing rural populations: ‘Capturing’ the grey nomad. Journal of Rural Studies, 27(2), 191–199. https://doi.org/10.1016/j.jrurstud.2011.01.004

Davies, A., & Dale, A. (1996). Locating the stranger rapist. Medicine, Science and the Law, 36(2), 146–156. https://doi.org/10.1177/002580249603600210

De Montjoye, Y.-A., Hidalgo, C. A., Verleysen, M., & Blondel, V. D. (2013). Unique in the crowd: The privacy bounds of human mobility. Scientific Reports, 3(1), 1–5. https://doi.org/10.1038/srep01376

Dietz, P. E., Hazelwood, R. R., & Warren, J. (1990). The sexually sadistic criminal and his offenses. The Bulletin of the American Academy of Psychiatry and the Law, 18(2), 163–178.

Duncan, T., Cohen, S. A., & Thulemark, M. (Eds.) (2013). Lifestyle mobilities: Intersections of travel, leisure and migration. Ashgate. https://doi.org/10.4324/9781315592404

Eck, J. (2003). Police problems: The complexity of problem theory, research and evaluation. Crime prevention studies, 15, 79–114.

Felson, M. (1987). Routine activities and crime prevention in the developing metropolis. Criminology, 25(4), 911–932. https://doi.org/10.1111/j.1745-9125.1987.tb00825.x

Felson, M. (2006). Crime and nature. Sage.

Felson, M., & Boba, R. L. (Eds.). (2010). Crime and everyday life (4th ed.). Sage.

Felson, M., & Boivin, R. (2015). Daily crime flows within a city. Crime Science, 4(1), 1–10. https://doi.org/10.1186/s40163-015-0039-0

Flamm, M., & Kaufmann, V. (2006). Operationalising the concept of motility: A qualitative study. Mobilities, 1(2), 167–189. https://doi.org/10.1080/17450100600726563

Garson, G. D. (2013). Weighted least squares regression. Statistical Associates Publishers.

Germann Molz, J. (2008). Global abode: Home and mobility in narratives of round-the-world travel. Space and Culture, 11(4), 325–342. https://doi.org/10.1177/1206331207308333

Gonzáles, M. C., Hidalgo, C. A., & Barabási, A.-L. (2008). Understanding individual human mobility patterns, Nature, 453(7196), 779–782. https://doi.org/10.1038/nature06958

Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime. Stanford University Press.

Groff, E. R., Taylor, R. B., Elesh, D. B., McGovern, J., & Johnson, L. (2014). Permeability across a metropolitan area: conceptualizing and operationalizing a macrolevel crime pattern theory. Environment and Planning A: Economy and Space, 46(1), 129–152. https://doi.org/10.1068/a45702

Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate data analysis: A global perspective (7th ed.). Pearson Education.

Hanson, R. K., & Bussière, M. T. (1998). Predicting relapse: A meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66(2), 348–362. https://doi.org/10.1037/0022-006X.66.2.348

Harris, A. J., Levenson, J. S., & Ackerman, A. R. (2014). Registered sex offenders in the United States: Behind the numbers. Crime and Delinquency, 60(1), 3–33. https://doi.org/10.1177/0011128712443179

Hawley, A. (1950). Human ecology: A theory of community structure. Ronald.

Heck, R. H., Thomas, S., & Tabata, L. (2012). Multilevel modeling of categorical outcomes using IBM SPSS. Routledge. https://doi.org/10.4324/9780203808986

Hollis, M. E., & Hankhouse, S. (2019). Crime risks and rural routines: A theoretical examination of guardianship activities in rural areas. International Journal of Rural Criminology, 4(2), 273–291. https://doi.org/10.18061/1811/87906

Kannisto, P. (2016). Extreme mobilities: Challenging the concept of ‘travel’. Annals of Tourism Research, 57, 220–233. https://doi.org/10.1016/j.annals.2016.01.005

Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. Guilford Press.

Khazanov, A. M. (1994). Nomads and the outside world (2nd ed.). University of Wisconsin Press.

Kinney, J. B., Brantingham, P. L., Wuschke, K., Kirk, M. G., & Brantingham, P. J. (2008). Crime attractors, generators and detractors: Land use and urban crime opportunities. Built environment, 34(1), 62–74. https://doi.org/10.2148/benv.34.1.62

Koth, B. A. (2013). Trans-Pacific bluewater sailors–Exemplar of a mobile lifestyle community. In T. Duncan, S. A. Cohen, & M. Thulemark (Eds.), Lifestyle mobilities: Intersections of travel, leisure and migration (pp. 143–158). Ashgate. https://doi.org/10.4324/9781315592404-10

Lammers, M., & Bernasco, W. (2013). Are mobile offenders less likely to be caught? The influence of the geographical dispersion of serial offenders’ crime locations on their probability of arrest. European Journal of Criminology, 10(2), 168–186. https://doi.org/10.1177/1477370812464533

Laub, J. H., & Sampson, R. J. (2001). Understanding desistance from crime. Crime and Justice, 28, 1–69. https://doi.org/10.1086/652208

Levenson, J. S. (2008). Collateral consequences of sex offender residence restrictions. Criminal Justice Studies, 21(2), 153–166. https://doi.org/10.1080/14786010802159822

Levenson, J. S. (2018). Hidden challenges: Sex offenders legislated into homelessness. Journal of Social Work, 18(3), 348–363. https://doi.org/10.1177/1468017316654811

Levenson, J. S., & Cotter, L. P. (2005a). The impact of sex offender residence restrictions: 1,000 feet from danger, or one step from absurd? International Journal of Offender Therapy and Comparative Criminology, 49(2), 168–178. https://doi.org/10.1177/0306624X04271304

Levenson, J. S., & Cotter, L. P. (2005b). The effect of Megan’s Law on sex offender reintegration. Journal of Contemporary Criminal Justice, 21(1), 49–66. https://doi.org/10.1177/1043986204271676

Levenson, J. S., & Vicencio, C. P. (2016). Residence restrictions. In E. L. Jeglic & C. Calkins (Eds.), Sexual violence: Evidence based policy and prevention (p. 51–65). Springer International Publishing. https://doi.org/10.1007/978-3-319-44504-5_4

Levenson, J. S., Ackerman, A. R., Socia, K. M., & Harris, A. J. (2015). Where for art thou? Transient sex offenders and residence restrictions. Criminal Justice Policy Review, 26(4), 319–344. https://doi.org/10.1177/0887403413512326

Levenson, J. S., Letourneau, E., Armstrong, K., & Zgoba, K. M. (2010). Failure to register as a sex offender: Is it associated with recidivism? Justice Quarterly, 27(3), 305–331. https://doi.org/10.1080/07418820902972399

Lussier, P., & Healey, J. (2009). Rediscovering Quetelet, again: The “aging” offender and the prediction of reoffending in a sample of adult sex offenders. Justice Quarterly, 26(4), 827–856. https://doi.org/10.1080/07418820802593360

Lussier, P., LeBlanc, M., & Proulx, J. (2005). The generality of criminal behavior: A confirmatory factor analysis of the criminal activity of sex offenders in adulthood. Journal of Criminal Justice, 33(2), 177–189. https://doi.org/10.1016/j.jcrimjus.2004.12.009

Mannering, F., Murakami, E., & Kim, S. G. (1994). Temporal stability of travelers' activity choice and home-stay duration: Some empirical evidence. Transportation, 21(4), 371–392. https://doi.org/10.1007/BF01110481

Marchetti, A.-M. (2002). Carceral impoverishment: Class inequality in the French penitentiary. Ethnography, 3(4), 416–434. https://doi.org/10.1177/1466138102003004014

Martineau, M., & Beauregard, E. (2016). Journey to murder: Examining the correlates of criminal mobility in sexual homicide. Police Practice and Research, 17(1), 68–83. https://doi.org/10.1080/15614263.2014.994215

Marvell, T. B., & Moody, C. E. (1998). The impact of out‐of‐state prison population on state homicide rates: Displacement and free‐rider effects. Criminology, 36(3), 513–536. https://doi.org/10.1111/j.1745-9125.1998.tb01257.x

Michaud, P. (2022). Should we stay or should we go? A critical review of journey-to-crime research [Manuscript submitted for publication]. School of Criminology, University of Montreal.

Mingus, W., & Burchfield, K. B. (2012). From prison to integration: Applying modified labeling theory to sex offenders. Criminal Justice Studies, 25(1), 97–109. https://doi.org/10.1080/1478601X.2012.657906

Morselli, C., & Royer, M.-N. (2008). Criminal mobility and criminal achievement. Journal of Research in Crime and Delinquency, 45(1), 4–21. https://doi.org/10.1177/0022427807309630

Murray A. T., Grubesic, T. H., Mack, E. A., Wei, R., Rey, S. J., Anselin, L., & Griffin, M. L. (2013). Convicted sex offender residential movements. In M. Leitner (Ed.), Geotechnologies and the environment: Crime modeling and mapping using geospatial technologies (Vol. 8, pp. 51–71). Springer. https://doi.org/10.1007/978-94-007-4997-9_3

Mustaine, E. E., Tewksbury, R., & Stengel, K. M. (2006). Residential location and mobility of registered sex offenders. American Journal of Criminal Justice, 30(2), 177–192. https://doi.org/10.1007/BF02885890

Muthén, B. O., & Satorra, A. (1995). Complex sample data in structural equation modeling. Sociological Methodology, 25, 267–316. https://doi.org/10.2307/271070

Nakagawa, S., Johnson, P. C., & Schielzeth, H. (2017). The coefficient of determination R2 and intra-class correlation coefficient from generalized linear mixed-effects models revisited and expanded. Journal of the Royal Society Interface, 14(134), 1–11. https://doi.org/10.1098/rsif.2017.0213

Newton, A., & Felson, M. (2015). Crime patterns in time and space: The dynamics of crime opportunities in urban areas. Crime Science, 4(1), 1–5. https://doi.org/10.1186/s40163-015-0025-6

O’Regan, M. (2013). Others have the clock but we have time: Alternative lifestyle mobilities and the resurgence of hitchhiking. In T. Duncan, S. A. Cohen, & M. Thulemark (Eds.), Lifestyle mobilities: Intersections of travel, leisure and migration (pp. 35–50). Ashgate. https://doi.org/10.4324/9781315592404-3

Pager, D. (2003). The mark of a criminal record. American Journal of Sociology, 108(5), 937–975. https://doi.org/10.1086/374403

Paik, L. (2017). Critical perspectives on intersectionality and criminology: Introduction. Theoretical Criminology, 21(1), 4–10. https://doi.org/10.1177/1362480616677495

Paquette, S., & Cortoni, F. (2021). Offense-supportive cognitions, atypical sexuality, problematic self-regulation, and perceived anonymity among online and contact sexual offenders against children. Archives of Sexual Behavior, 50(5), 2173–2187. https://doi.org/10.1007/s10508-020-01863-z

Pedersen, D. M. (1997). Psychological functions of privacy. Journal of Environmental Psychology, 17(2), 147–156. https://doi.org/10.1006/jevp.1997.0049

Peters, J. (2012, December 10). Was Israel Keyes the most meticulous serial killer of modern times? Slate. https://slate.com/news-and-politics/2012/12/israel-keyes-suicide-is-he-the-most-meticulous-serial-killer-of-modern-times.html

Piquero, A. R., Farrington, D. P., & Blumstein, A. (2003). The criminal career paradigm. Crime and Justice, 30, 359–506. https://doi.org/10.1086/652234

Quinsey, V. L., Lalumière, M. L., Rice, M. E., & Harris, G. T. (1995). Predicting sexual offenses. In J. C. Campbell (Ed.), Assessing dangerousness: Violence by sexual offenders, batterers, and child abusers (pp. 114–137). Sage.

Ratcliffe, J. H. (2006). A temporal constraint theory to explain opportunity-based spatial offending patterns. Journal of Research in Crime and Delinquency, 43(3), 261–291. https://doi.org/10.1177/0022427806286566

Rengert, G. F., & Wasilchick, J. (2000). Suburban burglary: A tale of two suburbs (2nd ed.). Charles C. Thomas.

Ressler, R. K., Burgess, A. W., & Douglas, J. E. (1988). Sexual homicide: Patterns and motives. The Free Press.

Reyes-García, V., & Pyhälä, A. (Eds.) (2017). Hunter-gatherers in a changing world. Springer. https://doi.org/10.1007/978-3-319-42271-8

Reynald, D. M. (2009). Guardianship in action: Developing a new tool for measurement. Crime Prevention and Community Safety, 11(1), 1–20. https://doi.org/10.1057/cpcs.2008.19

Reynald, D. M. (2010). Guardians on guardianship: Factors affecting the willingness to supervise, the ability to detect potential offenders, and the willingness to intervene. Journal of Research in Crime and Delinquency, 47(3), 358–390. https://doi.org/10.1177/0022427810365904

Rossmo, D. K. (1987). Fugitive migration patterns [Unpublished master's thesis]. Simon Fraser University.

Rossmo, D. K. (2000). Geographic profiling. CRC Press. https://doi.org/10.4324/9780367802011

Rydberg, J., Grommon, E., Huebner, B. M., & Bynum, T. (2014). The effect of statewide residency restrictions on sex offender post-release housing mobility. Justice Quarterly, 31(2), 421–444. https://doi.org/10.1080/07418825.2012.667141

Sampson, R. J., & Laub, J. H. (1995). Crime in the making: Pathways and turning points through life (Rev. ed.). Harvard University Press.

Savage, J., & Windsor, C. (2018). Sex offender residence restrictions and sex crimes against children: A comprehensive review. Aggression and Violent Behavior, 43, 13–25. https://doi.org/10.1016/j.avb.2018.08.002

Schnittker, J., & John, A. (2007). Enduring stigma: The long-term effects of incarceration on health. Journal of Health and Social Behavior, 48(2), 115–130. https://doi.org/10.1177/002214650704800202

Schwaner, S. L., Mcgaughey, D., & Tewksbury, R. (1998). Situational constraints and absconding behavior: Toward a typology of parole fugitives. Journal of Offender Rehabilitation, 27(1-2), 37–55. https://doi.org/10.1300/J076v27n01_04

Schwartz, A. (1973). Interpreting the effect of distance on migration. Journal of Political Economy, 81(5), 1153–1169. https://doi.org/10.1086/260111

Seto, M. C. (2008). Pedophilia and sexual offending against children: Theory, assessment, and intervention. American Psychological Association. https://doi.org/10.1037/11639-000

Socia, K. M. (2011). The policy implications of residence restrictions on sex offender housing in upstate NY. Criminology & Public Policy, 10(2), 351–389. https://doi.org/10.1111/j.1745-9133.2011.00713.x

Socia, K. M., Levenson, J. S., Ackerman, A. R., & Harris, A. J. (2015). “Brothers under the bridge”: Factors influencing the transience of registered sex offenders in Florida. Sexual Abuse, 27(6), 559–586. https://doi.org/10.1177/1079063214521472

Sommet, N., & Morselli, D. (2017). Keep calm and learn multilevel logistic modeling: A simplified three-step procedure using Stata, R, Mplus, and SPSS. International Review of Social Psychology, 30, 203–218. https://doi.org/10.5334/irsp.90

Song, C., Qu, Z., Blumm, N. & Barabási, A.-L. (2010). Limits of predictability in human mobility. Science, 327(5968), 1018–1021. https://doi.org/10.1126/science.1177170 

Spradley, J. P. (1970). You owe yourself a drunk: An ethnography of urban nomads. Little, Brown and Company.

Statistics Canada (1992). 1991 census area profiles (No. catalogue 95F0168X).

Statistics Canada (2007). 2006 community profiles (No. catalogue 92-591-XWE).

Statistics Canada (2012). Illustrated glossary, census year 2011 (No. catalogue 92-195-X).

Statistics Canada (2017). Municipalities in Canada with population decreases between 2011 and 2016 (No. catalogue 98-200-X2016002).

Terranova-Webb, A. (2010). “Getting down the road”: Understanding stable mobility in an American circus [Doctoral dissertation, The Open University]. https://doi.org/10.21954/ou.ro.0000ed8d

Tewksbury, R. (2002). Validity and utility of the Kentucky sex offender registry. Federal Probation, 66, 21–26.

Tewksbury, R. (2007). Exile at home: The unintended collateral consequences of sex offender residency restrictions. Harvard Civil Rights-Civil Liberties Law Review, 42, 531–540.

Tewksbury, R., & Lees, M. (2006). Perceptions of sex offender registration: Collateral consequences and community experiences. Sociological Spectrum, 26(3), 309–334. https://doi.org/10.1080/02732170500524246

Tewksbury, R., & Levenson, J. S. (2009). Stress experiences of family members of registered sex offenders. Behavioral Sciences & the Law, 27(4), 611–626. https://doi.org/10.1002/bsl.878

Thomas, T. (2011). The registration and monitoring of sex offenders: A comparative study. Routledge. https://doi.org/10.4324/9780203815472

Thompson, B. Y. (2019). The digital nomad lifestyle: (Remote) work/leisure balance, privilege, and constructed community. International Journal of the Sociology of Leisure, 2(1), 27–42. https://doi.org/10.1007/s41978-018-00030-y

Turley, E., & Hutzel, L. (2001). West Virginia sex offender study. Criminal Justice Statistical Analysis Center.

van Sleeuwen, S. E., Ruiter, S., & Steenbeek, W. (2021). Right place, right time? Making crime pattern theory time-specific. Crime Science, 10(1), 1–10. https://doi.org/10.1186/s40163-021-00139-8

Warren, J., Reboussin, R., Hazelwood, R. R., Cummings, A., Gibbs, N., & Trumbetta, S. (1998). Crime scene and distance correlates of serial rape. Journal of Quantitative Criminology, 14(1), 35–59. https://doi.org/10.1023/A:1023044408529

Zandbergen, P. A., & Hart T. C. (2006). Reducing housing options for convicted sex offenders: Investigating the impact of residency restriction laws using GIS. Justice Research and Policy, 8(2), 1–24. https://doi.org/10.3818/JRP.8.2.2006.1

Zuckerman, M. (2007). Sensation seeking and risky behavior. American Psychological Association. https://doi.org/10.1037/11555-000

Figure

Figure 1. Five components of the criminal-nomadism scale

Tables

Table 1. Descriptive statistics of the 12 independent variables

Independent categorical variables

n

%

White

Yes

400

89.3

No

48

10.7

Mother tongue is French

Yes

406

90.6

No

42

9.4

Type of sexual offenses

Exclusively against children

119

26.6

Exclusively against adolescents

85

19.0

Exclusively against adults

151

33.7

Mixed

89

19.9

Non-contact

4

0.9

Independent continuous variables

M

SD

Total number of criminal sentences

7.5

5.0

Number of years of schooling

8.6

3.1

Longest work experience (in months)

66.2

58.5

Longest intimate relationship (in months)

82.1

61.6

Age of onset — first crime

19.7

7.7

Duration of criminal career (in years)

16.2

9.3

Prison time (in years)

2.9

2.8

Specialization in sex offending

.36

.27

Total number of sex crime victims

2.9

2.3


Table 2. Descriptive statistics of the five components of the criminal-nomadism scale

n

%

Number of trips

0 trips

133

29.7

1-2 trips

149

33.3

3-4 trips

89

19.9

5-7 trips

48

10.7

8+ trips

29

6.5

Number of nodes

1 node

133

29.7

2 nodes

131

29.2

3 nodes

92

20.5

4 nodes

43

9.6

5+ nodes

49

10.9

Length of paths

0 km

133

29.7

1-249 km

135

30.1

250-999 km

101

22.5

1000-1999 km

44

9.8

2000+ km

35

7.8

Range

1-9 km

56

12.5

10-24 km

88

19.6

25-99 km

127

28.3

100-499 km

102

22.8

500+ km

75

16.7

Mesolevel activity space

0-49 km²

43

9.6

50-499 km²

96

21.4

500-1499 km²

199

44.4

1500-1999 km²

80

17.9

2000+ km²

30

6.7

Table 3. Results of hierarchical multiple linear regression analysis with weighted observations for several characteristics of men convicted of sex offenses predicting the score on the criminal-nomadism scale (N = 439)

Predictors

Criminal-nomadism scale

Model 1 (weighted)

R² = .346

Model 2 (weighted)

R² = .377

Model 3 (weighted)

R² = .402

Model 4 (weighted)

R² = .408

B

SE

Beta

B

SE

Beta

B

SE

Beta

B

SE

Beta

Total number of criminal sentences (log)

0.47

0.03

0.59***

0.45

0.03

0.57***

0.34

0.04

0.43***

0.34

0.06

0.43***

White (0/1)

0.14

0.06

0.09*

0.12

0.06

0.08*

0.13

0.06

0.09*

Mother tongue is French (0/1)

-0.32

0.09

-0.14**

-0.30

0.09

-0.13**

-0.30

0.09

-0.13**

Number of years of schooling

0.01

0.01

0.09*

0.01

0.01

0.09*

0.01

0.01

0.09*

Longest work experience (in months)

0.00

0.00

-0.05

0.00

0.00

-0.03

0.00

0.00

-0.02

Longest intimate relationship (in months)

0.00

0.00

0.00

0.00

0.00

0.03

0.00

0.00

0.02

Age of onset — first crime

0.00

0.00

-0.06

0.00

0.00

-0.08

Duration of criminal career (in years)

0.00

0.00

0.01

0.00

0.00

0.01

Prison time (in years)

0.05

0.01

0.19***

0.05

0.01

0.19***

Specialization in sex offending

0.01

0.12

0.01

Total number of sex crime victims

0.00

0.01

0.01

Sexual offenses exclusively against children (0/1)

0.07

0.06

0.07

Sexual offenses exclusively against adolescents (0/1)

0.13

0.07

0.09

Sexual offenses exclusively against adults (0/1)

0.06

0.06

0.05

*p < .05. **p < .01. ***p < .001

Table 4. Results of generalized linear mixed models with repeated-measure design predicting the odds of traveling, the odds of exploration and the distance traveled by men convicted of sex offenses between each sentencing occasion

Predictors

Model 1a

Travel or not? (0/1)

Model 2a

Explore or not? (0/1)

Model 3b

Distance traveled (km)

Coefficientc

SE

Coefficientc

SE

Coefficientc

SE

Fixed effectsd

Intercept (α00)

-0.52

0.27

0.66*

0.33

5.31***

0.19

Level 1: Sentencing occasions (SO)

 

 

 

 

 

 

Age at SO-A

-0.02*

0.01

-0.01

0.01

-0.02*

0.01

Progression (in %) of criminal career at SO-A

0.66*

0.27

-1.93***

0.35

0.39*

0.20

SO-A involved a sexual conviction (0/1)

-0.04

0.12

0.77***

0.19

0.04

0.07

Judicial municipality of SO-A has a population > 1,000,000 (0/1)

-0.73***

0.12

0.49*

0.19

-0.21***

0.06

Judicial municipality of SO-A has a population < 30,000 (0/1)

0.36*

0.16

0.18

0.17

0.16*

0.06

Days in prison between SO-A and SO-B (log)

0.07**

0.02

-0.03

0.03

0.00

0.01

Days of freedom between SO-A and SO-B (log)

0.14***

0.03

0.18***

0.05

0.04

0.02

Level 2: Individuals

 

 

 

 

 

 

White (0/1)

0.76**

0.26

-0.32

0.34

0.17

0.20

Mother tongue is French (0/1)

-0.49*

0.23

-0.11

0.24

-0.75***

0.18

Number of years of schooling

0.05*

0.02

0.04

0.02

0.02

0.02

Age of onset — first crime

-0.03*

0.01

0.04*

0.02

0.00

0.01

Random effectse

Repeated measures

Within-sentencing occasion variance

0.88***

0.03

0.96***

0.05

0.72***

0.07

Between-sentencing occasion covariance (rho)

0.29***

0.03

-0.10*

0.05

0.56***

0.05

Between-subject variance (µ0j)

0.70***

0.15

0.13

0.12

0.40***

0.09

Goodness of fit

-2 log-likelihood (unconditional mean model)

10852.96

4143.65

2609.92

-2 log-likelihood (conditional repeated-measure model)

10666.04

4279.76

2509.78

Fit improvement χ2 (d.f.)

186.92*** (13)

---

100.14*** (13)

N of sentencing occasions

2463

965

965

N of individuals

448

319

319

a Model uses a binomial distribution with a logit link (binary logistic regression).

b Model uses a gamma distribution with a log link (gamma regression).

c Robust covariances are presented.

d Continuous predictors were grand-mean centered.

e The repeated covariance uses a first-order autoregressive (AR1) structure and the random effect covariance uses a variance component structure.

*p < .05. **p < .01. ***p < .001 (two-tailed).

Comments
0
comment

No comments here

Why not start the discussion?