The relationship between population density and neighbourhood crime levels is a classic one found in criminological research. Urbanization has often been seen as detrimental to the preservation of social ties among neighbourhood residents. Early scholars such as Park (1929) were concerned by this trend and very pessimistic in their views. They considered the maintenance of social ties to be more difficult for residents in areas characterized by high population density because of the anonymity that is produced by high levels of population density in cities, thus setting the stage for crime and disorder.1 In this paper we ask the question to what extent population density, defined as the ratio of residents per unit of area, can be similarly seen as a structural background condition that sets the stage for ecological concentrations of crime and disorder in two different settings. Many scholars have addressed the question of why population density is related to crime and disorder from a theoretical point of view (Shaw and Mc Kay, 1942, Sampson and Groves, 1989). We know from numerous
studies that neighbourhood levels of crime tend to correlate with structural characteristics such as population density and disadvantage (poverty) at multiple levels of aggregation (Ouimet, 2000; Wikström, 1991). However, scholars seem to disagree in some respects: (1) regarding the effect of population density independent of structural disadvantage, (2) regarding the actual mechanisms at work in the relationship between population density and crime – amongst other things questions have been raised as to the role played by social trust in this relationship (Sampson, Raudenbush and Earls, 1997; Wikström and Dolmén, 2001) and, (3) there has also been some debate as to the independent role played by neighbourhood levels of disorder in the causation of neighbourhood levels of crime (Bursik and Grasmick, 1993; Sampson and Raudenbush, 1998). While the literature already contains a large body of empirical evidence relating to these questions, only very few scholars have addressed these issues from a comparative perspective. Studies are often restricted to a single setting. Sampson and Wikström (2008) have previously argued that more cross-national research is needed in order to
understand both how social structure and social mechanisms affect crime levels and how general the findings of ecological studies are across countries. Indeed, only very rarely have studies shown how the distribution of crime can be similarly explained in different countries (Sampson and Wikström, 2008). Sampson and Wikström employed identical measures of the key constructs and showed that similar processes were at work in explaining neighbourhood differences in Chicago and Stockholm.
The present study draws on such previous attempts to produce insights into the comparability of the relationship between population density and crime (Sampson & Wikström, 2008; Eisner & Wikström, 1999) but our effort is more modest since we do not have identical measures of our theoretical constructs. Thus this paper is not a comparative study in the strict sense of the term but rather an examination of a much debated criminological issue. Although we acknowledge that this may be
seen as a shortcoming, other attempts to compare different samples without perfect comparability in the measures employed have nonetheless contributed to the knowledge of how similar mechanisms shape crime and disorder in different settings (e.g. Svensson and Pauwels, 2008). Theories often have a middle-range status and thus allow for some level of generalisation. If a general theoretical model holds, the empirical relationships should be found in multiple settings and independently of the specific
measures used as indicators of the relevant constructs.2 Thus, this article contributes to the existing literature by examining the extent to which two theory-driven models of the relationship between population density and crime can be equally applied to data collected in two cities, Antwerp in Belgium and Malmö in Sweden. More specifically, we examine what happens to the relationship between population density and crime when neighbourhood disadvantage, social trust and disorder are introduced into the equations. In particular, we will focus on evaluating the strength of the relationships
between population density, disadvantage, social trust, disorder and crime, and in doing so we will be highlighting the “comparability” of results by pointing to similarities in the findings from the two cities.
This study is an aggregate-level study at the neighbourhood level. The study of crime at aggregate levels is necessary in order to improve our theorizing and understanding of the consequences of area structures for the unfolding of criminal events. It also contributes to the development of cross-level theories on the occurrence of crime (e.g. how area characteristics contribute to the selection of targets).