Max Planck Institute for Foreign and International Criminal Law

1. Introduction: About the Project

Generally, the European Crime and Safety Survey [ECSS] is looking at European‘s experiences with crime and crime prevention, and the police. Furthermore, it helps measuring attitudes towards crime and punishment by analysing data about personal experiences with selected offences of representative population samples from various European countries. Thus, the ECSS aims at developing a tool for measuring crime in various European countries. This project is financed with funds from the European Union‘s Sixth Framework Programme.

In a wider perspective, the project is embedded in the International Crime Victims Survey [ICVS], thus mainly applying the ICVS methodology. Actually, the ECSS is part of the current fifth sweep of the ICVS. The ICVS is itself the main item of an international comparative criminology project with standardised victim surveys. The ICVS was carried out in more than 70 countries all over the world during four completed sweeps (1989, 1992, 1996, and 2000). Since its beginning, the ICVS has been supported and promoted by a number of governmental and intergovernmental international institutions. The ICVS contains items on crime situations that cover a broad scope of possible victimisation experiences, which are particular for the modern urban context. Thus, the ICVS/ECSS provides a useful tool for the comparison of European crime rates and citizen‘s attitudes towards crime and punishment. In particular, the ICVS/ECSS serves three main aims: (1) Providing an alternative to police information on levels of crime; (2) Harnessing crime survey methodology for comparative purposes; (3) Extending information on who is most affected by crime.

2. Structure and Methodology of Our Analysis

For our first research report, we focused on the respondents‘ attitudes towards punishment and sentencing (punitiveness), represented in the survey questionnaire by two subsequent items. As regards our analysis methods, first we did a cross-country comparison with the German respondents as reference group. In a second step, we tried to explain the German respondents‘ punitiveness scores by using the predictive power of other variables in a categorical regression analysis with optimal scaling technique. After a thorough analysis and validation procedure of the regression model for the German total sample and different German subgroups, we tested the model on two additional contrast country samples, France and the UK, for a first external validation.

3. Description of the Results, Conclusions, and Outlook

As regards the general preference of a certain sentence category, the vast majority of all countries‘ respondents prefer the more lenient sentences “community service” and “fine” for the presented case scenario of a recidivist burglar. On the other hand, a total average of about one fourth of all countries‘ respondents selects the most severe sentence, an „unsuspended prison term”. Altogether, we found remarkable country differences with Germany ranking in the lower middle part of the analysed countries. For the further steps of the country comparison, we concentrated on the second questionnaire item, a follow-up question to specify the appropriate prison term. Over all countries, almost two-thirds of the respondents preferred rather short prison terms of “up to one year“. On the other hand, almost all country samples (except for France) covered the full range of available prison terms – actually, including “life sentence”! In the respective inferential statistic analyses, we found only the respondents from Denmark being significantly more lenient than the German sample, but six country samples with significantly higher values for punitiveness (UK, Estonia, Poland, Hungary, Ireland, and Portugal). Hence, Germany seems to have a rather low punitive profile as compared to other European countries.

Afterwards, we analysed the determinants of the German participants‘ punitiveness scores by using categorical regression analysis methods. On the basis of a review of the respective research literature, we selected a choice of altogether 17 items from the dataset as the pool of predictor variables for the initial model. After a systematic successive elimination process, we finally found a regression model that included nine predictor variables and that was able to explain 10.5 percent of the variance of the participants‘ punitiveness scores. The table below shows the German sample‘s final regression model‘s standard parameters as well as some further indicators for the model‘s reliability and validity.

Unsurprisingly, the majority of these variables were demographic or socio-economic indicators. Of all the nine predictor variables that were finally included in the regression model, the participants' age and sex were by far the most influential. Since these two "hard” demographic variables were dominating not only the general German model but also the models for the German subgroups and the contrast country models, they seem to be the "Big Two” for predicting punitiveness. From a methodological point of view, we first tested the internal validity of the regression model by analysing each of the nine-predictor variable's influence in a single-variable model. Furthermore, we also checked each variable's Delta-R-square as another indicator for the model's quality. As a further methodological check, we tested the model's stability on ten partial samples that were drawn from the total German sample. As a result of these analyses, three of the predictors (sex, age, and occupational status) were perfectly reliable through all test runs. Four others (victimisation status, general life satisfaction, household size, and fear of crime) were sufficiently reliable. The influence of the two weakest predictor variables (level of self-security and income) was not stable enough to trust the respective results without caution. Later, the model was also tested and further validated on subgroups for the variables sex and victimisation status, as these variables had proven to be the main sources of influence for the respondents' punitiveness.

In the final step of analysis for this report, we adjusted the regression model to two other countries, which contrasted (on average) maximally regarding the punitiveness scores, i.e., France, as the least punitive, and the United Kingdom, as the most punitive country. We found some fundamental and rather influential country differences, which are probably based on essential historical and cultural differences in the perception of crime and punishment. This is, in our opinion, one of the most interesting results of international victim survey research. Consequently, a subsequent research report on the ECSS data (in preparation) will mainly focus on such basic differences (as well as similarities) between the European countries. There, we are going to present a country typology that is based on a cluster analysis of various ECSS variables and that is subsequently validated using different country indicators from international statistics regarding crime, justice, the police, economy, and education.