![Sample our Law journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5942/TOC-Subject-Sample-2021-Law.jpg"})
Abstract
Immigrants are known to be overrepresented in the crime statistics of Nordic countries. However, the composition of immigrant populations varies across countries both in terms of immigrants’ country of origin as well as their population structure (age and sex). Cross-country comparison of crime rates is always difficult because of differences in legal systems, but it is even more challenging when using very broad categories of immigrants, lumping heterogeneous groups together. Previous studies have largely compared the entire immigrant population of a country with the majority population, which under-appreciates the heterogeneity that exists across immigrant groups. In this paper, we compare the crime rates in Norway and Finland, while adding additional nuances by reporting crime rates for 25 specific immigrant groups relative to the majority population. The data are gathered from Finnish and Norwegian administrative records, representing or comprising the resident population aged 15–64. We analyse both violent crime and property crime, and we present the results adjusted for population structure (sex and age). The results show considerable similarity in the rank order of crime rates of immigrant groups in the two Nordic countries. Although the current study is mainly descriptive, it aims to set some limits to what it is to be explained. In fact, the diversity is so great that it is questionable whether one should treat immigrants as one single group at all. Whatever the causal mechanisms driving immigrant crime are, it seems plausible that some similar processes are operating across the Nordic region.
Acknowledgements
We are grateful for the useful comments on the draft from Lars Østby and Venla Salmi.
Notes
1 Email: [email protected].
2 Email: [email protected].
3 The official Danish crime statistics also provide highly disaggregated figures, but these are not adjusted for population structure.
4 Our definition of Eastern Europe includes Albania, Belarus, Bulgaria, Czech Republic, Hungary, Latvia, Lithuania, Moldova, Romania, Slovakia and Ukraine, while Western Europe includes immigrants from Austria, Belgium, Denmark, Faroe Islands, France, Greece, Iceland, Ireland, Italy, Luxembourg, Malta, the Netherlands, Portugal, Spain and Switzerland.
5 A key strength of the case-control design is its ability to study outcomes that are rare on population level with relatively small samples (Woodward, Citation2005). In this design, the case (crime outcome = 1) and control (crime outcome = 0) groups are sampled with unequal inclusion probabilities from the corresponding population groups. While different sampling fractions lead to an invalid estimation of relative risk or odds, the odds ratio can be used as the metric for effect size, as it is unaffected by the sampling fractions (see footnote 2).
6 If we assume that total population figures for the majority of the population are 4,500,000 (no crime) and 500,000 (crime), and the corresponding figures for immigrants are 400,000 and 100,000, the relative risk of crime for immigrants is (100/500)/(500/5,000) = 2.00, and the corresponding odds ratio is (100/400)/(500/4,500) = 2.25. If we take a 10% sample of the non-offender population and a 50% sample of the offender population, the estimate of relative risk is impacted by the different sampling fractions (50/90)/(250/700) = 1.56. The same applies to estimates of probability and odds. The ratio of two odds, however, remains the same, as (50/40)/(250/450) = 2.25.
7 It is easy to overlook that although the averages might be similar across countries, there might be additional important differences in the distributions.
8 The importance of socioeconomic differences for explaining immigrants’ overrepresentation is much discussed in public debates as well as in academic studies (see Hällsten et al., Citation2013). We have also shown in Table that income levels differ in the immigration groups in this study. There is, therefore, general interest in results when controlling for socioeconomic differences. For reasons explained above, controlling for income, for example, is problematic and is not reported as a part of the main results. We, nevertheless, ran the above models with controls for income (Appendix 1). Personal income is a strong correlate of violent crime, especially of larceny, in both countries. In models for violence, adjusting for income attenuates the coefficients further when compared to M2 in Table , but most groups with a higher level of violence than the majority population retain a significant positive association in both Norway and Finland (13 groups). In larceny models, on the contrary, the association between immigrant background and property crime remains positive and significant only for former Yugoslavian countries in Norway. What is more, 17 groups have a significantly lower risk of property crime than the majority population after controls. In Finland, nine countries or country groups have a significant positive association after income controls. Although this change might seem spectacular, the interpretation is not entirely clear, as personal income is clearly endogenous to crime. These results show that several of the groups do not have a higher crime rate compared with those with similar earnings in the majority population; importantly, though, this is not necessarily related to income per se.