Routine Activities: A Cross-National Assessment of a Criminological Perspective

The routine activity approach to the etiology of crime has gained considerable attention in the last few years. Its propositions have been tested using a variety of data sources and analysis methods. To date, however, the majority of analyses have used data collected within one nation and have employed unidimensional indicators. This article explores the macrostructural tenets of the approach based upon a sample of 52 nations spanning a 25-year period (1960-1984). The findings offer qualified support for the approach and uncover interesting anomalies. First, the model appears to be crimespecific, applying more to property crime than personal crime. Second, the best fitting model is nonlinear and specifies threshold effects. These findings are discussed in light of current research on the routine activity approach. Over the last ten years there has been a explosion in research and a rapid expansion of knowledge concerning a criminological approach variously called routine activity, lifestyle, or opportunity theory.' Because of the relative newness of this approach, the literature is vague on the specific relationships among (1) social structural conditions and routine activities or opportunities and (2) those routine activities or opportunities and the risk of victimization or crime.2 Two prominent models emerge from a perusal of the empirical literature, models that are conceptually very similar but differ by what is empirically investigated and what is assumed a priori. Each model views crime or the risk of victimization as a process whereby social structural change causes a change in the nature and frequency of routine activities and, subsequently, in the levels of risk. However, while one model assumes a specific social structure (e.g., proportion of single-person households, percent of women in the workforce, and amount of leisure time) and then empirically investigates the effect of routine activities on risk (cf. Garofalo 1987; Lynch 1987; Maxfield 1987; Miethe, Stafford & Long 1987), the other model investigates the empirical relationship between social structure and risk while assuming the intervening routine The author wishes to thank James P. Lynch, Michael G. Maxfield, and Sandra Baxterfor their assistance in conceptual development, Peter Basiotis for his aid in data analysis, and Mona Danner for her aid in data management. Direct correspondence to the Department of Justice, Law, and Society, The American University, Washington, DC 20016. ? The University of North Carolina Press Social Forces, September 1991, 70(1):147-163 This content downloaded from 207.46.13.159 on Sun, 23 Oct 2016 05:16:43 UTC All use subject to http://about.jstor.org/terms 148 / Social Forces 70:1, September 1991 activities (i.e., not measuring or testing them within the model) (cf. Cantor & Land 1985; Cohen & Felson 1979). In an attempt to advance our understanding of the efficacy of the routine activity approach in explaining the risk of crime, this article (1) investigates the second and less researched model, where the effect of structural change on crime rates is investigated while assuming the mediating or intervening effect of routine activities, (2) employs a cross-national sample of highly divergent social structures,3 (3) investigates the relative and simultaneous effects of the routine activity approach's central concepts, and (4) devises multiple indicators to measure more precisely the multidimensional nature of the approach's central concepts. In addition, following the tradition of Cohen and Felson (1979), this model is applied to both personal and property crime rates. Elements of the Model Although activity models in the literature incorporate varying explanatory concepts as well as levels of analyses, each includes reference to a central core of three concepts. This core includes consideration of the (1) suitability of the target, (2) proximity of the victim to a pool of motivated offenders, and (3) level of guardianship over the target. The present research relies on this literature and develops a three-part composite model for testing with cross-national data.