Sonia Stefanizzi and Giancarlo Manzi
Those against immigration often argue that “immigration is bad for safety” or “immigration means more crime”. Let us state emphatically: there is no direct link between immigration and crime. In a recent meta-analysis on the relationship between immigration and crime rates, Ousey & Kubrin found that there is a weak negative immigration-crime association, with homicide being the most negatively significant crime associated with immigration. Adelmann et al., for their part, investigated 40 years of an immigration-crime relationship in metropolitan areas in the U.S., and found that immigration is consistently linked to decreases in violent and property crime.
Opinion surveys in Western countries suggest that people are increasingly against immigration. In the UK, according to European Social Survey (ESS) data, the percentage of people against allowing immigrants from poorer countries has risen from around 15% in 2002 to around 22% in 2014. However, results from the same survey seem to confirm that more immigration does not mean more crime, as the percentage of people declaring to have been victim of burglary or assault in the last five years decreased from around 28% in 2002 to around 18% in 2014.
In Europe, starting from the economic crisis in 2008, a seemingly unstoppable wave of populism has pervaded the continent. According to electionresources.org, parties like the Party for Freedom (PVV) in the Netherlands, the Five Star Movement (M5S) in Italy, the UK Independence party (UKIP) in the UK or the Alternative for Germany party (AFD) in Germany have passed from one-digit preference percentages in pre-2008 parliamentary elections to two-digit figures at the times of the refugee crisis. Parties like the Hungarian Civic Alliance (FIDESZ) in Hungary have maintained percentages close to 50%, governing with semi-xenophobic policies of closures against each possible concession to immigration, often in contrast with the recommendations of the European Union. The equation “more immigration equals more crime” has always been one of the favourite propagandistic slogans of these parties. So is it that, to cite a recent Observer quote, populism festers when those [populist] leaders decide that those they represent are too stupid to grapple with real-world complexity?
The search for a complete security where individuals are no longer able to count on the strength of the bonds of community, but have more and more a duty to be responsible for themselves (1) gives rise to demands for borders, walls, and scapegoats. In the perception of many European citizens, migrants are representative of a nightmare. They express the precariousness and fragility of the human condition. In a certain sense they represent the possibility of being “superfluous”, which we ourselves, because of increasing precarity economic, could easily become. Immigrants have become the main bearers of what the domestic population fears and against which it draws borders. Borders, paradoxically, simply amplify the very differences they seek to contain. Precisely because of this, borders and the differences they create become self-legitimating (2).
In other words, even poverty, in an era of globalization, becomes strategically functional to the market, because it represents, so to speak, the living proof of what it means to be free from uncertainty, so that the sight of the poor constrains the non-poor from imagining a different world (3). The poor are often criminalized, along with foreigners, according to the rites of the scapegoat (4). Sociality, which characterizes our societies, is expressed, sometimes in orgies of compassion and charity, sometimes in outbursts of immense aggression against a newly discovered public enemy (3). The collective anxiety, in waiting to find a tangible threat against which to manifest itself, is mobilized against any enemy and, often, the stranger is identified tout-court with the criminal that threatens the personal safety of citizens. Politicians tend exactly to exploit this hardship for political ends.
This has nowadays become a very sensitive topic and extreme care should be taken when considering the relation of populism and the increase in the “immigration fear”. Official opinion surveys like the Eurobarometer, or the ESS, seldom have explicit questions about this relationship in their questionnaires, even in special editions. With these surveys one can combine similar questions (“Are you against immigration?” “Do you think in your country crime is increasing?”), and obtain composite indicators about this relationship. Private survey questionnaires might have questions on this, but then one cannot entirely rely on their estimates, as incomplete information is provided on technical aspects of the surveys. In other words, we cannot completely rely on surveys about the relationship immigration-insecurity. Moreover, in responding to sensitive topics people tend to hide their real feelings. Answers given to objective questions are often in contradiction with respect to answers to perception/opinion questions relating to the same topic.
We have to take into consideration multiple sources of information about this relationship, giving them a different weight, according to their reliability, and end up with a weighted average of the estimates. This is sometimes called “blended” or “combined” estimation, and in this era of a big data revolution includes also considering complementary data sources and social networks like Twitter. Twitter estimates can be obtained using text mining techniques. In the case of the immigration-crime relationship, automatic classification of tweets between favourable and unfavourable should be accompanied by manual evaluation when neutral tweets are retrieved or there is training of the automatic coder in order to limit false positive responses. Then, Bayesian analysis can be used to directly model the bias contained in each estimate source with the subjective evaluation of the researcher between more and less reliable estimates (for details on these methods see Lohr and Brick).
Preliminary results conducted by the authors using this methodology shows that “official” surveys tend to give generally less unfavourable attitudes about the immigration-crime relationship. On the other side, Twitter and private survey estimates present the most unfavourable results.
ESS 2014 data were used to obtain an index of the perceived immigration-insecurity association (PII index) by combining a perceived safety index constructed with the responses to two questions in the ESS 2014 questionnaire regarding the victimization with respect burglary/assault in the last 5 years and the feeling about walking alone in area of residence after dark, and an index of the attitude towards migrants constructed from 6 questions in the questionnaire (“Should your country allow people from abroad to come and live here?”, “Would you say it is generally bad or good for your country’s economy that people come and live here from other countries?”, etc.).
The ESS 2014 PII index resulted 34.2% for Spain, 39.9% for France, 36.5% for the UK and 38.8% for Italy. Ipsos 2014 data were used to construct the PII index by considering the responses to two questions very related to the II relationship, i.e. “Would you say that immigration has generally had a positive or negative impact on your country?, and “Immigration is causing my country to change in ways that I don’t like”, and taking a simple average of the proportion of the unfavourable answers. The Ipsos 2014 PII index resulted 60.0% for Spain, 71.0% for France, 60.5% for the UK and 77.0% for Italy. Finally, for each of the four countries 1,000 tweets regarding the immigration-insecurity relationship were analysed and classified in “good” or “bad” tweets and the proportion of the bad tweets were regarded as the Twitter 2014 PII index, which resulted 59.4% for Spain, 60.7% for France, 60.5% for the UK and 75.3% for Italy.
The Bayesian analysis combines these estimates and gives estimates, which are more or less close to the original estimates according to the belief about the bias in the original sources. For example, if one accepts that 45% is the true PII index (therefore accepting that ESS is the less biased source as it gives estimates closer to this value), then the combined PII index will result 36.9% for Spain, 49.0% for France, 38.4% for the UK and 51.1% for Italy. The true hypothesized estimate included exogenously in the model can be taken from a previous census or a more reliable source. Other sources like Eurobarometer can be added in the model.
In conclusion, here we face the well-known “folk devils” phenomenon mentioned by Cohen (5), i.e. subjects considered to be enemies of public order, identified as figures such as homeless people, vagabonds, prostitutes or migrants who are seen as representing the source of moral and are panic skilfully nurtured by the media and populist parties. However, people react differently when responding to an official survey questionnaire or tweeting spontaneously. In the first case they exert self-control, whereas in the second case they tend to remove their inhibitory brakes. With sensitive topics such as the immigration-insecurity relationship one cannot rely on one source of information only. Bayesian analysis allows for the consideration of multiple sources of information, and mediates within a range of outputs enforcing the reliability of the estimates.
(1) Zygmunt Bauman, ‘Liquid Life’. 2005, Cambridge (UK): Polity Press.
(2) Frederick Barth, ‘Introduction’. In Barth F. (Ed.) ‘Ethnic Groups and Boundaries. The Social Organization of Culture Difference’, 1969, Long Grove (IL): Waveland Press, pp. 9-38.
(3) Zygmunt Bauman, Society Under Siege, 2002, Cambridge (UK): Polity Press.
(4) Réne Girard, ‘Job, the Victim of His People’, 1987, Stanford (CA): Stanford University Press.
(5) Cohen, S. (1972). Folk Devils and Moral Panics: The Construction of the Mods and Rockers. London: MacGibbon & Kee.
Sonia Stefanizzi is Professor of Sociology, Department of Sociology and Social Research, University of Milan-Bicocca, Italy. Giancarlo Manzi is Senior Lecturer in Statistics, Department of Economics, Management and Quantitative Methods, University of Milan, Italy.