Does the mobility of foreign offenders fit the general pattern of mobility?

Distance (decay)

In the first half of the 20th century, White (1932) demonstrated that offenders make little effort to travel, carrying out their crimes at locations near their homes. The findings of his pioneering work were confirmed later that century by several others (see, for example, Capone and Nichols, 1975; Hesseling, 1992; Phillips, 1980; Reppetto, 1974; Rhodes and Conly, 1981). More recently, scholars have extended this finding to reveal the existence of a ‘distance decay’ function. This function prescribes that most offences are committed near the home address of the offender and that the number of crimes diminishes as the distance from home increases (for example, Bernasco, 2006; Canter and Youngs, 2008a; Phillips, 1980; Rattner and Portnov, 2007; Rengert et al., 1999; Turner, 1969; Van Koppen and Jansen, 1998).

Despite the general observation of the distance decay pattern, there has been a debate about whether this pattern is observed only at the aggregate level or at the level of the individual offender too. Although this discussion was rather abstract at first (Rengert et al., 1999; Van Koppen and De Keijser, 1997), recent empirical research has shown that a large amount of the variation in offending patterns exists between offenders and only to a lesser extent within offenders (Smith et al., 2009; Townsley and Sidebottom, 2010; Van Daele, 2010).

Anchor points

In order to calculate a journey to crime, information about the starting point and the terminus are needed. Information about the terminus, or crime site, can be found for the most part in official data. But the starting point of the criminal’s home address has been assumed. Some have elevated this to the status of a principle and named it ‘domocentricity’ (Canter and Gregory, 1994; Canter and Larkin, 1993; Sarangi and Youngs, 2006). The centrality of the home address enables researchers to successfully develop a geographical profile. With this technique, the location of the crime scenes is used to derive an offender’s most likely activity space or starting area. This often refers to the home area (Canter, 2003; Canter and Youngs, 2008a; Kocsis et al., 2002; Rossmo, 1995, 2000; Rossmo and Rombouts, 2008). Wiles and Costello (2000) observed that offenders regularly start their crime trips at other locations – friends’ homes, the workplace or leisure venues. It is thus also demonstrated empirically that a crime trip does not always begin at home but may start from bars, shops, friends’ homes, bus stations or schools.

Direction

Rengert refers to direction as ‘heading towards certain areas’ (Rengert, 2004: 171–2). Pyle et al. (1974) consider city centres to be outstanding opportunity structures. Notification of criminal opportunities depends on people’s daily routines. Both routine activity theory (Cohen and Felson, 1979; Felson and Cohen, 1980) and pattern theory stress the importance of criminal opportunities (Brantingham and Brantingham, 1981, 1993b). The routines of potential victims create opportunity structures for motivated offenders, and the everyday activities of offenders enable them to exploit these opportunities successfully.

Most of these findings have originated from research on local offenders: offenders with a residence and awareness of space within the target area. Yet this is probably not the case for all offenders. Some will have settled only recently in a particular area and will have yet to develop extensive environmental knowledge about the setting. Bernasco (2010) found, for instance, that residential history plays an important role in criminal location choice. Offenders who have recently moved are still looking for criminal opportunities in the vicinity of their previous residence. Consequently, offenders who are not permanent residents may be unaware of offending opportunities in that area. Our research examines the applicability of offender mobility principles on such a set of offenders. In particular, we pay attention to foreign offenders who come from Eastern Europe.

We propose to test three hypotheses: (1) the pattern of foreign offenders follows a distance decay curve; (2) foreign offenders start from their temporary residence; (3) foreign offenders, just like indigenous offenders, tend to travel toward criminal opportunity structures.

Police data

Our quantitative approach uses police data. This approach connects the offender with the place of residence and the crime location, enabling us to calculate distances. The quantitative data set contains all serious property crimes, that is, property crimes with aggravating circumstances such as violence or intrusion, in Belgium for the years 2002–2006.

Because mobility is offender related, only crimes with known offenders were selected (which is roughly 10 percent of the total number of crimes). Countries in Eastern Europe are located between 800 and 2000 km from Belgium. It is unlikely that daily crime trips are undertaken over such a distance. To avoid large error margins because of vague or mistaken information about residence, only recorded residences in Belgium are considered (these may be temporary). This results in a data set of over 67,000 crime–offender combinations or crime trips (N = 67,981: 51,385 by Belgian offenders, 7078 by East European offenders and 9518 by other foreigners). As such, the main unit of analysis is offender–offence combinations or crime trips, which has been used by Hodgson and Costello (2006) too. A substantial part of the database consists of multiple offenders (31 percent). Hence, the database has a layered structure. The intra-class correlation coefficient (ICC) of .55 indicates that there is a resemblance of crime trips within individual offenders (Hox, 2010). As suggested by Townsley and Sidebottom (2010), this requires a multilevel approach. ¹

Given the nature of police data, crimes and residences are geocoded on the basis of the municipality in which they are located. The data set contains both residences and crime locations, making it possible to calculate Euclidian crime distances. For crimes that end in a different region from where they started, the distance between the centroids of both regions has been used. In cases where both the residence and the crime are located in the same municipality, we follow the method used by Bernasco (2006) and calculate the distance as half of the square root of the area (Ghosh, 1951). There are 589 municipalities in Belgium with an average area of 51.8 km² (Min = 1.1; Max = 213.8, SD = 37.8).

In cases with co-offending, the trip for each offender was calculated individually. Thus, if two offenders co-offend, two trips will be calculated: one from each residence to the crime location. Although both offenders may have met before at one of these residences, the data do not allow us to identify this actual starting point. Furthermore, relying only on this actual starting point would neglect the travelled distance and the spatial knowledge of the other offender involved; hence our decision to treat them as separate crime trips.

The data set contains both crime (date, location) and offender (nationality, age) variables, but also variables with information on the locations (municipality coordinates, affluence index, population density). The strength of this method lies in the fact that a large sample – both in actual size and in covered geographical area – is used and the data may therefore be quantitatively analysed. This method allows us to calculate the distances covered and the distance decay pattern of East European offenders (Hypothesis 1), the percentages of recorded anchor points (Hypothesis 2) and the main locations of offence targets (Hypothesis 3). The chief weakness of this method is that only information on apprehended offenders is incorporated and no in-depth information on the crime trip can be provided. One may analyse the distance between the municipalities, but that does not mean that offenders actually started from home or used the shortest route.

Case files

To compensate for these weaknesses in the data from police statistics, we added a twofold qualitative approach to the research and applied grounded theory principles. The first is a case files analysis. We analysed 26 case files concerning East European offenders. These case files were obtained from five distinct judicial districts that are quite diverse in location (from West to East), geography (both border regions and inland districts), size (smaller and larger areas) and building density (rural and urban areas). Given this diversity, these five districts function as a sampling frame within which a selection of cases was made, based on permissions and the availability and involvement of East European offenders. ² The files refer to cases from the period 2000–7. Some offenders committed their crimes in a time span of three months, whereas others were active criminals for nearly three years (33 months). The offenders were responsible for a variety of property crimes, including residential burglaries, commercial burglaries and robberies, metal thefts and ram raids. The case files enable us to assess not only the target location but also the starting point – which may not always coincide with the offender’s residence. Both in the choice of police districts and in the choice of actual case files, our analysis has sought a maximum degree of heterogeneity. Because of the great diversity, however, the variety of cases helps us to understand the differences in travel to crime places. This method is mainly used to find out where these offenders started their crime trips, in other words, where their anchor points lie (Hypothesis 2).

Offender interviews

In addition, we interviewed 21 Romanian offenders in prison in 2009. Criminals from Romania are, together with people from Albania and the former Yugoslavia, the main group of East European offenders in Belgium (FOD Justitie, FOD Kanselarij van de Eerste Minister, and FOD Binnenlandse zaken, 2007). The interviews were conducted with the help of an interpreter. This gave the respondents the possibility to answer questions in their mother tongue. Only convicted offenders were approached, because these individuals have less reason to lie or conceal information about the crimes they have committed. Those interviewed were identified from a selection of suitable respondents (N = 67) in Belgian prisons who were asked to participate in the research. The potential sample size was greatly reduced because of: (1) the limitation of interviewing only convicted offenders (not those in custody); (2) the fact that many of the offenders were in Belgium illegally and were therefore expelled after serving part of their punishment; (3) the prison administration data were not up to date; and (4) potential respondents refused to cooperate. The interviews developed around a topic list, which is presented as Appendix 1. The interviews dealt with four main issues: general introductory questions, topics concerning travelling and living routines, involvement in crime and criminal location choice.

Interviewing offenders of the same nationality and in the same language allowed us to have the same interpreter, creating a high level of uniformity throughout the interviews. It was made clear that the interview would deal with both their general background information and their criminal activities. The interviews were conducted in one of the prisons’ separate rooms (rooms most often used for prisoners to meet their lawyers) with only the respondent, the interpreter and the researcher present. Most interviews took between 60 and 90 minutes. The respondents were asked whether the interview could be tape-recorded. If this was not the case, notes were taken.

Through face-to-face, semi-structured interviews with offenders, first-hand information was obtained on starting points, target choices and directional motivations, including space awareness. These additional data were used in particular for assessing offenders’ target location choice (Hypothesis 3).

The respondents were all male, between 22 and 45 years of age. Most of them were in their twenties (10 people) or thirties (9 people). None had long-term employment. Most had worked for only a couple of days a week and on a temporary and illegal basis (moonlighting). They had worked mostly in the construction industry, horticulture, the construction of exhibition stands, selling cars, selling flowers, or begging.

Case file analysis

The case file analysis came up with similar results. Some offenders undertook short crime trips, but the majority of crime trips took them further away. In all the case files the offenders had a large operating area that extended beyond the city borders and to over 50 km. In 15 case files the offenders did not offend in the city where they stayed overnight and were therefore always mobile.

Although it is difficult to uncover the exact route that was followed, some offenders used their mobile phones during crime trips. Tracing this activity might link offenders to certain crimes and routes. These cases indicated that offenders use the motorway to reach their targets. It is quite remarkable that they travel such distances without even considering closer targets.

Several types of anchor point were observed. In only 3 cases was there a secondary anchor point that functioned as a convergence setting for criminals. No similar information was available for the other 23 cases. However, police investigations put most emphasis on offenders, not places. The chances are, therefore, that in a number of cases a convergence setting did exist but it may have escaped police attention because of their different focus and because the police are able to build up a picture of the offenders even without assessing the convergence setting.

The residences that are encountered in the files are quite fixed. In only 1 case did the offenders regularly move to other areas. In all other cases, the residences were stationary. In 18 cases the premises were fixed, and in the other 7 cases offenders did move but still remained in the same city or region. This basically implies that the anchor points can be regarded as stationary: given the large distances they covered during crime trips, the residential changes had little to no effect on the direction and length of the trips. Furthermore, crime travelling in some cases could be observed through mobile phone tracing. In 1 case, the offenders travelled back and forth during the night from a city in the south of Belgium to the coast area (over 150 km one way) to commit a couple of burglaries. Because they regularly called each other on the phone, police forces were able to trace their travel route and time quite easily. Their speed and location during their calls mean there is no other conclusion than that they used the motorway to travel.

In the vast majority of the case files (18), the offenders departed from large cities. In 5 cases, the offenders stayed in the border region, operating on both sides. In only 3 cases did the offenders live elsewhere.

Most offenders develop spatial knowledge in ways that directly correspond to criminal activity: they go out scouting beforehand (15 cases); they operate in the same neighbourhoods, targeting the same houses as they did before or nearby houses (7 cases) – so-called near-repeat burglaries (Bernasco, 2008; Townsley et al., 2003); or they may even use maps to learn about the area (2 cases). However, in only 2 cases was information gathered through activities that were completely independent from crime.

Offender interviews

As with the case file analysis, various distance patterns were observed: 6 interviewed offenders operated exclusively near their anchor point and may therefore be regarded as local offenders; the other 15 respondents were more mobile and covered larger areas. Of the latter, 10 did not operate at all in the area where they stayed overnight.

During the interviews, 6 offenders said they travelled around randomly and started looking for targets relatively close to their starting point, but ended up further away because they found nothing suitable nearby: 2 of them commented on their surprise when they realized that they had travelled such distances; the other 4 did not consider their distances to be so great.

I wasted a fortune on gasoline . . . I started to look around [in nearby municipalities], but then a neighbour appeared or something . . . . A bit further, a roll-down shutter was brought up and so on . . . By the time I found a target, I had driven a lot. (Respondent b2)

Only 2 respondents had lived in more rural areas; the others all resided in one of Belgium’s major cities. Most interviewed offenders (15) started their crime trips from the places where they also stayed overnight. Seldom was this known as their official residence. Often, they had spent the night with several others in small apartments owned by slum landlords (10 respondents). Others (6) had managed to rent a temporary apartment for themselves. A third type of anchor point was found for offenders who stayed overnight in less common places that were not mentioned in the official crime statistics: hotels (6), railway stations (2), abandoned buildings (1) or a car (1). These places may evolve over time: people who first stayed with others might later be able to rent an apartment by themselves (2 respondents), and people sleeping in hotels might alternatively sleep at a friend’s place or elsewhere (2 respondents).

Not always the same. First a couple of days in [city 1], then a couple of days in [city 2], and then back. (Respondent b14)

In several cases, other places than home – in its broad interpretation as a sleeping place – functioned as anchor points. People met in pubs (7), railway stations (2) and other public places (4). Here, they made arrangements for future crimes and often initiated their crime trips. Some bars merely functioned as meeting places that provided sufficient anonymity. In other cases, pub owners were actively involved in selling stolen goods or even taking a coordinating role in crime.

Mostly we divided the loot . . . and then about an hour later back to the bar . . . and we talked to the owner of the pub to make a deal concerning the price. He bought just about everything. (Respondent a2)

Some criminal organizations use non-profit organizations and their headquarters as a front for crime. Doing this has the advantage that people can walk in and out without looking suspicious and, compared with a regular pub, the non-profit organization is able to keep out most unwanted visitors.

We found that 7 offenders had no fixed anchor point. They alternated crime trips starting from home with crime trips that departed from certain convergence settings. This hampers a clear assessment of the crime trip, because one cannot easily say which anchor point was used for which crimes.

Crimes outside the anchor point area may occur in other cities, but the most popular locations were wealthy rural areas. The offenders interviewed gave three main reasons for travelling outside the cities. First, they were in a particular place to meet old friends or to search for work (6 respondents). In this case, they had non-criminal motives for travelling (it was the first time they had gone there). Hence, they still had little awareness about the area.

I went with a friend from [city1] to [town], because he suggested having a drink with someone he knew who had lived there for a while. . . . Then we met him at the pub, because he had a wife and child waiting at home and we didn’t want to get him involved . . . and we walked a bit and started burgling. (Respondent b10)

Second, they indicated that rural areas afforded ‘quiet’, in the sense of a better place, to ‘work’ (3 respondents). Instead of the crowded but rather anonymous cities, they chose to operate in regions where they expected to be seldom confronted by any possible guardians. The offenders explained that, for burglary, they particularly looked for houses that are secluded, so that their visibility would be low. In urban environments, these types of targets are rare. A third reason the respondents mentioned is the affluence of rural environments. Offenders indicated that they went to richer districts and that they often also took into account signs of wealth when choosing their actual targets (3 respondents). Some offenders (6) were not able to give a particular reason for their selection of target area. They drove around and operated randomly (5 persons) or acted together with someone else who made the decisions (1 respondent).

Given the mobility of the offenders in our sample, we are also interested in their space awareness. It turns out that they had few daily routines. They had only limited access to work and leisure activities and no extensively developed space awareness. If they knew the region in which they operated, it was often because they had already committed other crimes in that region or because they had performed some reconnaissance, both of which can hardly be described as regular developments of spatial awareness.